There may be some typos in this chapter. Please notify if you find any.

CHAPTER 3
RISER DESIGN AND PLACEMENT
3-1 General remarks. We will now show how the theory of Chapter 2 and supplementary material can
be applied to the actual quantitative design of risers for complex castings.
Risering is, of course, a process designed to prevent the formation of shrinkage voids in the casting upon
solidification. To indicate the proportion of solidification shrinkage in the total contraction of a casting,
let us review the relevant figures (in percent) for plain carbon steel during the three stages of contraction
illustrated in Fig. 3–1:
(a) During the liquid state (contraction in the liquid metal), approximately 1.6% by volume per 100°F
(56°C);
(b) during the transformation from the liquid to the solid state, about 3.0% by volume (values for other
alloys are given in Table 3-1);
(c) during the solid state, about 7.2% by volume, from the solidification range to 70°F (39°C).
The point to recognize is that contraction (b) can take place at constant temperature or over a narrow range
and is the result of a density change accompanying the transformation from liquid to solid. Visualize a
one-inch cube of liquid steel at the solidification temperature, as shown Fig. 3-1.
A thin outer shell will freeze at constant temperature, and there will then be entrapped within the fixed
outer dimensions a volume of liquid whose subsequent transformation to the denser solid will produce
shrinkage of type (b). In castings of irregular sections, it is also possible for the shrinkage of stage (a) to
add to the voids if the gates are thin and freeze off before the casting reaches the freezing range. Note that
shrinkage voids are not produced by contraction of type (c) which occurs in the solidified metal. (This
type can, however, result in serious stresses, tears, and cracks, as discussed in Chapter 6).
Table 3-1
The problem at hand, shrinkage porosity, can be solved
by controlling the solidification pattern and thermal
gradients, so that the voids are produced outside the
body of the casting proper. It is evident that to develop
the best engineering properties in a given casting, the
shrinkage voids must either be eliminated or isolated at
a location of low stress.
We shall see that to produce a sound casting, the riser
or reservoir of liquid metal which is to compensate for
the shrinkage must satisfy two independent
requirements.
(a) Riser size. If the riser is to supply liquid metal to feed
the casting shrinkage, it must freeze after the casting
(except in certain gray irons discussed at the end of this
chapter). When the mold material surrounding the riser and the casting is the same, then the ratio
(volume/area)2 of the riser must exceed that of the casting (unless an external source of heat such as an
exothermic compound or electric arc is placed over the riser, or the casting is chilled).

Figure 3-1. Development of shrinkage void in a cubic casting. (a) Liquid. (b) Liquid + solid. (c) Liquid +
solid. (d) Solid.
(b) Riser placement. For an alloy with high centerline feeding resistance, a casting will require a closer
spacing of risers than for other alloys. In other words, the effective feeding distance of a riser in a "wide
freezing-band alloy” is smaller than in a "narrow-band alloy."
It is evident that because of the differences among alloys in solidification patterns, the excess of riser-
freezing time over casting-freezing time and the variation in riser placement may vary widely. However,
the fundamental ideas which have been evolved in Chapter 2 provide a good background for developing
and applying the experimental data.
We shall proceed now to discuss the construction of several graphs used to calculate the riser size and
then take up the calculation of riser placement. Since most of the published work to date has dealt with
steel and ductile iron, the discussions will necessarily be limited to these materials. Special attention to
gray iron, brass, bronze, and aluminum is given, however, at the end of the chapter.
3-2 Risering curves for steel. Riser size is determined by two factors: first, the freezing time of the riser
must exceed that of the casting, at least to some extent, and second, the riser must supply sufficient feed
metal to compensate for the liquid-to-solid shrinkage. Caine [1] has evaluated these requirements as shown
by the risering curve of Fig. 3–2(a). The curve in Fig. 3-2 (a) does not include any safety factor, which
can be allowed for by increasing the constant c in the equation by whatever amount is deemed advisable.
For example, a factor of 5% should be enough for steel. In the figure, the x-axis represents the ratio of the
freezing time of the riser to that of the casting, which, as mentioned above, must always be greater than
1:1 to obtain a sound casting. As an index of freezing time, Caine uses the ratio (surface area/volume),
which is inversely related to the V2/A2 ratio of Chvorinov, developed in Chapter 2. Caine's reasoning is
that heat dissipation is a function of the surface area of the casting, while the heat content is a function of
volume. It is assumed intuitively that the linear relation A/V determines cooling rate and hence is inversely
related to freezing time. While the Chvorinov relationship has a firmer theoretical basis, the simpler Caine
ratio apparently falls within the limit of error, considering the many other factors involved.
The y-axis represents the volume ratio of riser to casting. Here it is postulated that the riser will be required
to furnish feed metal to compensate at least for the liquid-to-solid shrinkage, which is approximately 3%
by volume for steel. The riser size, therefore, must be greater than 3% of the casting volume. This limiting
case of high yield, i.e., large casting/riser ratio, is reached only for extremely thin plates of very large
surface/volume ratio.
For example, a plate of dimensions 10 × 10 × 0.1 in. has an A/V-ratio of
100 + 100 + 4(10 × 0.1)
= 20.4
10 × 10 × 0.1
This volume would require 0.3 in3 of feed metal. A cube riser of 0.5 in. would have the ratio
𝐴 0.5 × 0.5 × 6 1.5
= = = 12,
𝑉 0.5 × 0.5 × 0.5 0.125

Fig. 3-2. (a) Risering curve for 0.3% C steel castings and the basic risering equation used to calculate riser
dimensions [1].
Fig. 3-2. (b) Risering curve for ductile iron (total carbon 3.6%, silicon 2.5%) [15]. Cross latched region is
the most efficient area for risering to complete soundness; side blind risers, clamped green- sand molds
gated into risers.
or a much slower cooling rate than the plate. However, the amount of feed metal, 0.125 in3, would be
insufficient even if it were available. For this reason, castings with high surface area/volume ratios require
larger risers than called for by cooling-rate considerations alone, as shown by the right-hand portion of
the graph in Fig. 3–2.
On the other hand, as the casting becomes compact and its cooling rate decreases, there is always adequate
feed metal in the riser, since, to cool at a sufficiently slow rate, the riser must be both compact and larger
than the casting in all dimensions. Consider, for example, the feeding of a 4-in. cube:
𝐴 4 × 4 × 6 96
= = = 1.5
𝑉 4 × 4 × 4 64
A cylindrical riser 4.5 in. in diameter and 4.5 in. high will be needed to produce a sound casting, as shown
by the following calculation. Using the most compact cylinder (diameter = height), so that A/V = 6/d=
1.33, we find that its A/V-ratio is
𝐴/𝑉𝑐𝑎𝑠𝑡𝑖𝑛𝑔 1.5
= = 1.13,
𝐴/𝑉𝑟𝑖𝑠𝑒𝑟 1.33
and its volume ratio is
𝑉𝑟𝑖𝑠𝑒𝑟 71.5
= = 1.11
𝑉𝑐𝑎𝑠𝑡𝑖𝑛𝑔 64.0

These coordinates satisfy the graph of Fig. 3-2. Improvement in yield can be obtained by the use of chills,
which can increase the effective surface area of cooling by a factor of five. Additional risering-curve data
(for ductile cast iron) are given in Fig. 3–2(b).
3–3 Naval Research Laboratory method of riser calculation. After several years of successful use of
the Caine risering calculation, the NRL group [2] worked out a new and simplified procedure (illustrated
in Figs. 3-3, 3-4, and 3-5), which has the advantage of eliminating trial-and-error calculations.

Fig. 3–3. Risering curve for steel (approximately 0.2% C to 0.5% C) [2].
The method is based on the observation that the ratio (riser volume/casting volume) must be greater for
chunky castings because of their relatively low surface/volume ratios. (This requirement is also a
characteristic of the Caine relationship). A major simplification is obtained, however, by the device of
introducing the shape factor (L + W)/T to replace the tedious calculation of surface/volume ratios for
complex castings. The length L, width W, and thickness T are computed by using the maximum
dimensions of the parent section of the casting. For example, if a casting has a number of small
appendages, then these are not used in the calculation of the shape factor; however, they do enter into the
calculation of the casting volume in a special way, as we shall discuss in Section 3-7.
The NRL method may be simply illustrated by the following practical example. The most economical
riser size for a steel casting of dimensions 5 × 10 × 2 in. is calculated by first computing the shape factor
(L + W)/T, which for this casting is (10 + 5)/2 = 7.5. Then, from Fig. 3-3, the ratio (riser volume/casting
volume), or Rv/Cv, is found to be 0.55. Since the volume of this casting is 100 in3, the riser volume required
is 0.55 × 100 in3 = 55 in3. From Fig. 3-4 it is seen that this volume is provided by a riser 4.5 in. in diameter
and 3.5 in. high.
However, it can be shown empirically that the most economical h/d ratio for risers attached to the side of
the casting is unity. On the other hand, the most economical h/d ratio for top risers is one-half, the lower
curve on each of the charts. Therefore, if the casting under consideration is to be fed by a side riser, the
best riser dimensions are: diameter 4.25 in. and height 4.25 in. If, however, a top riser is employed, it
should be 5.25 in. in diameter and 2.63 in. high or, more practically, 5.0 in. in diameter and 2.75 in. high.

Fig. 3-4. Chart for conversion of required riser volume to riser dimensions.
Fig. 3–5. Naval Research Laboratory risering chart for irregular sections. (a) Percentages of parasitic
volume to be added to parent volume to determine riser volume.
For a circular plate 10 in. in diameter and 2 in. thick, the (L + W)/T shape factor is (10 + 10)/2 = 10, but
the actual volume of 157 in3 is used in the volume calculation for the proper amount of feed metal. For a
cylinder 4 in. in diameter and 10 in. long, (L + W)/T = (10 + 4)/4= 3.5, and again the actual volume is
employed.
It has been demonstrated in practice that both the NRL and the Caine curves give approximately the same
riser dimensions for simple shapes. The NRL method, however, has been further extended to include
calculations for complex shapes such as those shown in Fig. 3–6. When ribs or other appendages are thin,
they do not appreciably increase the freezing time of the main portion of the casting, and therefore only a
small increase of liquid metal is needed in the riser to allow for the appendages. As the appendage becomes
heavier, the riser must be increased considerably to ensure that the freezing time of the riser is sufficiently
longer than that of the casting. Very thin fins can be used to reduce the cooling time; however, the effect
of such an arrangement is difficult to calculate and is not taken into account in Fig. 3–5.
The calculation of additional riser volume in terms of the percentage of liquid metal to be added to the
riser volume is indicated in Fig. 3-5. All appendages can be considered as approximations of bars or plates.
(A casting is considered to be a plate when the width of its cross section is greater than three times the
height. When the width is less than three times the height, the casting is considered to be a bar). Because
of the greater surface area, a bar will have a greater cooling rate than a plate of the same thickness. This
is evident in the case of a plate feeding a bar: a plate 1 in. thick is equivalent to a bar 1.6 in. thick. At this
size both have the same cooling rate, and the bar must be treated as part of the parent casting. However,
when a plate fed by a bar reaches 60% of the bar thickness, then it must no longer be treated as parasitic.
To illustrate the calculation, let us assume that a projection of cross section 1 in² and 4 in. long is added
to the plate of dimensions 5 × 10 × 2 in. discussed earlier. This addition does not enter into the calculation
of the (L + W)/T factor, which remains at 7.5. The volume of the parasitic bar is 4 in3, and from Fig. 3-5
it is evident that 30% of the parasite volume, or 1.2 in3, should be added to the parent volume. A riser
volume of 55.6 in3 is now required.
Fig. 3–6. Examples of casting conditions of different shapes and thicknesses [8]. (a) Bar-plate. (b) Plate-
plate. (c) Bar-bar. (d) Plate-bar.
Hollow cylindrical shapes such as bushings also present a special case. The heat flow from the center core
is restricted, and the casting as a whole has a lower cooling rate than a plate of the same cross section. The
simplest approach is to consider the shape as a plate, but to correct for an "effective plate thickness” (as
indicated by experimental observations) as follows.
Let T be the true wall thickness and Te the effective plate thickness. Then the following approximations
may be used:

As an illustration, let us consider the computation for a bushing whose dimensions are: outside diameter
12 in., inside diameter 4 in., and height 12 in. From the above table, the thickness of an equivalent plate
is 4.56 in. The shape factor is
8𝜋(𝑐𝑖𝑟𝑐𝑢𝑚𝑓𝑒𝑟𝑒𝑛𝑐𝑒) + 12
= 8.1,
4.56
𝑅𝑣
= 0.47,
𝐶𝑣
and the actual casting and riser volumes are
Cv = (62 – 22)12π = 1205 in3,
 Rv = 567 in3
Hence, according to the chart (Fig. 3–5), the riser should be 1112 in, in diameter and 512 in. high.

3–4 Feeding distance. All of the preceding calculations are based upon the assumption that the riser or
risers are placed within the effective feeding distances of the casting sections requiring liquid metal. For
example, from the curve it is possible to calculate the dimensions of a riser designed to feed a bar 1 in. in
diameter and 21 in. long, but obviously, because of the long feeding path, the resulting section, having
only a single riser at one end, would not be sound. Let us now proceed to calculate effective feeding
distances from the available experimental data. It will be recalled from the discussions of Chapter 2 that
each combination of alloy and mold material exhibits its own characteristic centerline feeding resistance.
It is necessary to determine experimentally the effective feeding distance for each combination until
sufficient data are compiled to permit the derivation of a general relationship. The most detailed
experimental results for cast steel have been published by the Naval Research Laboratory [3,4].
In the determination of feeding distance, it is assumed, for simplicity, that any casting can be divided into
plate, bar, and cubical (or spherical) sections. The cubical or spherical sections offer no problem of feeding
distance because the riser can be placed near the location to be fed. Hence, if we can develop feeding
distance data for bar and plate sections, we shall be able to position risers properly for any casting.
The plate and bar sections require individual attention because in the plate we encounter dendritic growth
proceeding from two principal walls, while in the bar four walls are involved. It is necessary, then, to pour
liquid metal and observe the feeding distances from risers in bars and in plates of different thicknesses.
(a) Feeding distance in bar sections. The NRL data for 4 × 4 in. bars of different lengths are presented in
Fig. 3–7. Two types of observations were made: first, temperature measurements were performed at
various centerline locations in the bars during cooling, and then the bars were inspected radiographically.
In all cases the radiographic inspection disclosed a sound section near the riser and another at the far end
of the bar. The 12-in. bar was completely sound, but both the 16-in. and the 24-in. bars exhibited centerline
shrinkage. There was some scatter in the results, as shown by the overlapping of the crosshatched areas in
the figure. It is interesting to note that the shrinkage zone approached the riser more closely than the far
end of the bar. The thermal data explain this observation quite convincingly.
The thermocouple readings are then used to develop a plot of temperature distribution at selected time
intervals. This method is similar to that employed in graphs of solidification (Chapter 2). For clarity, a
few selected lines from Fig. 3–7 (b) are reproduced in Fig. 3–8. In the feeding of a given centerline
location, the thermal gradient is most important for the period of time during which the last stage of
solidification (solidus) occurs. After this time interval has passed, nothing can be done to fill in shrinkage
voids which may have resulted during solidification. During the last stage of solidification, the dendritic
growth in the area, and hence the centerline feeding resistance, will be at a maximum.
To determine the thermal gradient at a given station during the last stage, a tangent is drawn to the
particular curve of distance versus temperature which passes through the solidus at that station. Note that
the gradient for each station is determined at a different elapsed time but åt the same temperature (the
solidus). By comparing the thermal-gradient data derived in this way with the soundness data, one can see
that in all cases a thermal gradient of greater than 6°F/in. is required to allow proper feeding. The 16- and
24-in. bars exhibit gradients of 0 to 2°F/in. at the center regions, indicating that these sections complete
their freezing, and so block off feed metal at about the same time.
The gradient data also explain why the sound region encountered at the end of the bar away from the riser
is longer than that at the riser. The faster cooling rate provided by the end face causes this section to freeze
well in advance of the center portion, thus drawing feed metal from the center. This metal, of course, is
replaced by riser metal. The cooling effect of the end face provides a greater thermal gradient over a
greater distance than does the riser.
After experimenting with bars from 2 in. to 8 in. in cross section, the Naval Research Laboratory derived
the following general relationship for bars in this range: effective feeding distance 𝐷𝑚𝑎𝑥 = 6√𝑇, where
T is the bar thickness and D is measured from the edge of the riser.

Figure 3-7(a), (b). Solidification of steel bars of cross section 4in. [3].
Figure 3-7(c). Solidification of steel bars of Fig. 3–8. Calculation of thermal gradient
cross section 4 in. [3]. from Fig. 3–7(b).

When the bar is of greater length than the feeding distance of a riser, it is necessary to use additional risers.
It should be emphasized that the feeding distance of a riser can be applied in any direction. For example,
in the 4 × 4 in. bar already discussed, if a riser 8 in. in diameter were centrally located, a sound 32-in. bar
could be produced (12-in. feeding distance in each direction from the riser edge). If a still greater length
is required, it is necessary to place two risers on the bar. In this case, the feeding distance for metal between
the risers is not 6√𝑇 from each riser edge, but is 1.2T because there is no cold end face between the risers,
Fig. 3-9.
The individual contributions of the riser and of the cold face at the end of the bar to the production of
sound regions are shown in Fig. 3–9. These contributions add up to the maximum length of sound bar that
can be obtained; for example, in a 4-in. bar, the riser contribution is 1.2T, that of the edge is 1.8T, and
hence the total contribution is 3T or 12 in.
Fig. 3–9. Feeding relationships in bars [8].
(b) Feeding distance in plate sections. The feeding distances
in plates have been established by a series of experiments [4]
similar to those for bars, using plates of thickness 12 in. to 4
in. As predicted, a given riser has a greater feeding distance
in a plate section because dendritic growth occurs only from
two principal walls, and therefore less resistance is offered to
the transport of feed metal. A thermal gradient as low as
1°F/in. can be tolerated for a horizontal distance equal to the
thickness of the plate, in comparison to the requirement of
6°F/in. minimum for bars. The feeding distance for plates 1
to 4 in. in thickness is 4.5T when both riser and cold end-wall
gradients are taken into account. The riser gradient prevails
for a shorter portion of the total distance (2T) than the end-
wall gradient (2.5T), just as for bars. Therefore, when the
feeding distance between risers is calculated, each riser feeds
a distance equal to 2T, not 4.5T. These relationships are
summarized in Fig. 3-10.

Fig. 3-10. Summary of end wall and riser gradients in plates [8].
 (c) Feeding distance in complex sections. In calculating riser size, it was necessary to develop
relationships among combinations of bars and plates (parent versus parasite), and the feeding distance
problem requires similar attention. When a heavy plate joins a light plate for its full width (Fig. 3-11), the
heavy plate is more effective than a riser. This follows from the fact that the riser has only a limited
attachment, at its periphery, to the plate, whereas the heavy plate feeds throughout its length. By a series
of systematic experiments, it was established [8] that the feeding distance in the thin plate is 3.5TH, where
TH is the thickness of the heavy section. Therefore a 1 in. plate can be fed by an adjoining 4-in. plate
section for 14 in. instead of for only 4.5 in.
The feeding distance of a riser over the heavy section, however, is reduced to DH = (TH – TL) + 4.5, where
DH is the feeding distance of the riser in the heavy section and TH and TL are the thicknesses of the heavy
and light sections, respectively. More elaborate modifications have been developed for intermediate
sections, for which we find the relation DM = 3.5TH - TL, where DM is the feeding distance in the middle
section. When a very light section, e.g., a 18-in. plate, is attached to a heavy(3-in.) section, the feeding
distance in the heavy section is not reduced, since the light section accelerates rather than slows down the
end-plate cooling.

Fig. 3-11. Feeding distances in varying plate sections [8].

The case of a light section bar attached perpendicularly to a heavy plate has not been investigated, but it
may be expected that the feeding distance in the bar will be the same as for a well-risered bar.
3–5 Other effects of complex sections and designs. From the many examples already discussed, it should
be apparent that any alteration in a section will lead to a change in cooling rate. Such alterations are of
two general types: (a) changes in direction (L-shaped versus straight bar) and (b) presence of intersections.

Fig. 3–12. Effect of change in direction, uniform section [5].

Fig. 3–13. T, X, V, and other intersections [5]. (Regions marked by crosses are "hot spots.")
(a) Effects of changes in direction. One effect of a change in direction of the uniform section illustrated in
Fig. 3-12 is the decrease in the volume of sand per unit of surface area which occurs on the side where the
included angle is less than 180°. When a condition of this type is encountered, the casting may develop
shrinkage because of the retarded cooling, i.e., the metal on both sides of the hot spot tends to freeze and
wall off the zone from feed metal.
Other difficulties, such as hot tearing and development of residual stresses, are also introduced. These will
be discussed in Chapter 6.
(b) Effects of intersections. Intersections provide a striking case of retarded cooling rate, as illustrated in
Fig. 3-13. There are three ways of coping with this problem.
(i) It is often effective to chill the side of the section where the re-entrant angle is located, although this
technique increases the molding cost. Also, the chilling may be overdone, and there is danger of choking
off the feed metal so that hot tearing may result at the edge of the chill.
(ii) Redesigning to lighten the metal section at the point of inter-section is often more desirable than
method (i). (Full details are given in reference 5).
(iii) Another possibility is to mold a thin fin across the valley of the intersection. This increases the surface
area of the metal and therefore the cooling rate.
3-6 Effect of chills. It was pointed out in Chapter 2 that the substitution of metal chills for all or a portion
of the sand-mold surface has two pronounced effects. Because the cooling rate of the casting is increased,
the start and end portions of the freezing waves are closer together in moving from the surface to the center
of the casting. This means that the mushy-liquid + dendritic-solid region is narrowed and the centerline
feeding resistance is reduced.
Chills, therefore, have two distinct functions in producing sound castings. First, the feeding distance in a
given riser-casting combination is increased because the centerline feeding resistance is lowered.
Unfortunately, no data are yet available to illustrate this effect quantitatively. One outstanding example,
however, may be cited. The Griffin Wheel Laboratories have developed a process for making a railroad
car wheel (33 in. in diameter) of 0.7% carbon steel, in a graphite mold. The entire rim section (100 in. in
circumference), as well as most of the plate or web section, is fed from four risers, so that a feeding
distance of over 12 in. per riser is indicated. Since the rim can be approximated by a 2-in. bar, we have
here a feeding distance of 6T as compared with the distance 2T expected in a sand mold. This difference
arises from the decrease in centerline feeding resistance for 0.6% carbon steel which drops from 54% to
19% (sand versus chilled) as calculated in Chapter 2.
Before proceeding further, we must emphasize that the effects of chills just discussed were for uniformly
chilled castings. It is found that even with uniform chilling, which causes more rapid freezing than sand
does, the freezing distance is improved in almost all cases. Only when the thermal conductivity of the
metal is much greater than that of the chill, e.g., for copper, is this method of chilling less effective.
Chills are far more commonly used to obtain a steeper thermal gradient in the longitudinal direction. In a
plate or bar casting where it is necessary to place several risers on top of the section (Fig. 3-14), the feeding
distance of the risers may be markedly increased and their number reduced by chilling. Consider first a
chill placed at the end of a plate or bar section. Its effect on feeding distance is relatively minor, since a
strong gradient is already present, caused by the sand-metal interface.
Figure 3-14. Combination of chills and risers for maximum feeding distance [8].
Accordingly, the feeding distance is increased only by a factor of T (the section thickness) for a bar, and
by a factor of 2T for a plate, (Fig. 3-14).
By contrast, let us consider the effect of a chill placed between risers. [With merely a chill in the drag
(bottom) surface, the effect upon the thermal gradient is the same as if an end chill were placed in the
mold (Fig. 3-14)]. The feeding distance of each riser is now 4.5T + 2 in., and for a bar, it is 6√𝑇 + 𝑇.
Thus, for example, in a 1-in. plate, the distance between risers can now be 13 in. instead of 4 in. This
effect is of obvious importance in all alloys when large plate sections, such as ball-mill liners, are to be
fed by a number of risers. In the selection of chills for this purpose, we are guided by the results of thermal
studies showing that chills of thickness and width equal to the plate are adequate. No increase in cooling
rate during the critical solidification period is obtained with water-cooled copper chills or heavier metal
chills.
3–7 Application of risering principles to complex castings. To illustrate the calculation of riser size and
placement let us consider the casting of a bearing housing shown in Fig. 3–15(a). We should recognize
first of all that this casting is made up essentially of two plates, the base and the cylindrical section
(unrolled). Secondly, it will not be possible to feed this casting with one riser because of the feeding
distance involved. Hence we decide to riser the casting from two locations as shown in Fig. 3–15(b).
The base and half the cylinder may be considered as a plate 11 × 5.5 × 1 in. with an appendage 8 × 5 × 1
in. The shape factor is
11 + 5.5
= 16.5,
1
whence we obtain a risering volume factor of 0.25. The volume of this shape is 105 in3, and therefore a
riser 3.5 in. in diameter and 2.7 in. high is required. Note that the volume of the base is taken at full value
because it is as thick as the cylinder. The half cylinder and boss away from the base have the same shape
factor (16.5). The volume of the boss is taken at full value which, combined with the half cylinder, equals
72 in3. Thus, a riser 3 in. in diameter and 2.6 in. high is needed.
Since the feeding distance of the risers is not sufficient, chills are employed as shown in the sketch. The
feeding distances from the riser to the end of the base and to the bottom of the cylindrical section are close
to the maximum, and a pilot casting should be made to determine whether additional chills are needed.

Figure 3-15.
3-8 Risering of gray iron. The risering of gray iron has been reserved for special attention, since it is a
complex problem having some features in common with steel, whereas others are completely different.
According to the thermal-analysis data given in Chapter 2, solidification of a 3% carbon gray iron takes
place by two distinct mechanisms:
(1) Liquidus to eutectic; separation of austenitic dendrites.
(2). Eutectic; concurrent precipitation of graphite and austenite.
Reaction (1) is similar to that for steel, i.e., the precipitation of solid austenite involves the formation of a
denser phase than that of the liquid, and therefore feed metal is required from a riser at this time. Also, as
shown by the thermal analysis data of Chapter 2, reaction (1) practically reaches completion before the
inception of reaction (2). During reaction (1), then, liquid is transported from the riser to the casting.
On the other hand, reaction (2) involves an expansion, since the density of the solid eutectic mixture is
less than that of the parent liquid. During this reaction, metal flows back into the riser, a process known
as "purging." [6]. If the riser is small and has frozen over at the top, great pressure will be generated within
the casting and the walls will actually be bowed outward. However, if the riser is open to the atmosphere,
the pressure is relieved by metal flow. This can be of major importance in determining dimensional
accuracy.
The relative shrinkage due to reaction (1) and the expansion of reaction (2) cancel each other [7] when
about 1.5% graphitic carbon is precipitated in the eutectic. This carbon value is related to the solubility of
carbon in the austenite, which in turn is affected by silicon. The percentage of eutectic graphite can be
expressed as
Eutectic graphite = total carbon – 2 + 0.1(%Si).
When this value is equal to 1.5% carbon or more, no risering is required. Below this value risers are needed
in proportion to the decrease in the percentage of eutectic graphite. Gray iron is, therefore, the only
common exception to the rule that the riser must freeze after the casting. The riser need stay liquid only
long enough to meet part of the early casting demand for metal to compensate for the liquid → austenite
reaction. At the end of this reaction, the eutectic reaction will produce an expansion and will "purge” part
of this riser metal back to the risers.
3-9 Risering of brass, aluminum, and magnesium. A search of the cast-metals literature will disclose
fully as many articles dealing with these nonferrous materials as with steel. Why then has so much space
been given here to steel? The key lies in the fact that the data for steel (and also for ductile iron and gray
iron) are quantitative, whereas the other data are largely qualitative and descriptive.
We may, however, predict something about the feeding behavior of a number of other alloys from the
centerline feeding-resistance data of Chapter 2. In sand molds we expect that alloys with a centerline
feeding resistance similar to that of steel will have similar risering characteristics, whereas those with high
resistance will be more difficult to feed. We can predict relationships like those shown in Table 3-2.
Recent work at the Naval Research Laboratory indicates that these predictions are correct. The feeding
distance for manganese bronze (similar to 60-40 brass) in plates is 5.5T versus 4.5T for steel, where T is
the thickness of the plate. Furthermore, in 88-10-2 bronze, shrinkage is encountered along the full length
of the bar.
Table 3-2
In general, then, it appears that the risering
data for steel may be applied to all alloys
whose centerline feeding resistance is lower
than that of steel. For alloys with great
resistance, local chilling is advisable at those
portions of the casting for which maximum
strength is indicated. For example, it has
been shown for 85-5-5-5 bronze that for 2 ×
2 in. bars and plates 1 in. thick, chilling is
required to produce a gradient of over
60°F/in., which is the minimum gradient
that will ensure soundness in this material
[10].

REFERENCES
1. J. B. CAINE, "Risering Castings,” Trans. A.F.S. 57, 66–76 (1949).
2. H. F. BISHOP, E. T. MYSKOWSKI, and W. S. PELLINI, "A Simplified Method for Determining Riser
Dimensions," Trans. A.F.S. 63, 271-281 (1955).
3. H. F. BISHOP, E. T. MYSKOWSKI, and W. S. PELLINI, “Soundness of Cast Steel Bars," Trans.
A.F.S. 59, 174 (1951).
4. H. F. BISHOP and W. S. PELLINI, "The Contribution of Riser and Chill Edge Effects to Soundness of
Cast Steel Plates," Trans. A.F.S. 58, 185–197 (1950).
5. Steel Casting Handbook, Steel Founders Society of America, Cleveland, O., 1950.
6. R. P. DUNPHY and W. S. PELLINI, "A Solidification Dilatometer and Its Application to Gray Iron,”
Trans. A.F.S. 60, 783–788 (1952)
7. W. A. SCHMIDT, E. SULLIVAN, and H. F. TAYLOR, "Risering of Gray Iron Castings," Trans. A.F.S.
62, 76 (1954).
8. W. S. PELLINI, "Factors Which Determine Riser Adequacy and Feeding Range," Trans. A.F.S. 61, 61-
80 (1953).
9. H. F. BISHOP and W. H. JOHNSON, The Foundry, Vol. 84, No. 2, p. 70 (1956); No. 3, p. 136 (1956).
10. R. A. FLINN and C. F. MIELKE, "Pressure Tightness of 85-5-5-5 Bronze, Thermal Gradients During
Solidification," Trans. A.F.S. 67, 385–392 (1959).

GENERAL REFERENCES
Transactions A.F.S., The Foundry, and R. W. RUDDLE, "Risering of Castings," in The Running and
Gating of Castings. The Institute of Metals Monograph and Report Series 19 (1956).

PROBLEMS
1. Using (a) the Caine method, (b) the NRL method, design a risering system, taking into account ease of
riser removal and molding procedure, that will produce sound castings with maximum yield for the
following simple shapes:
(i) a plate of dimensions 10 × 10 × 20 in.,
(ii) a cube of dimensions 6 × 6 × 6 in.,
(iii) a ball-mill liner of dimensions 50 × 60 × 2 in.,
(iv) a pipe 30 in. long; outside diameter 5 in., inside diameter 3 in.
2. (a) Calculate an economical risering system that will ensure soundness in the casting (0.30% steel)
shown in Fig. 3–16.

Figure 3-16
(b) Calculate an economical risering system for the same casting, assuming that the density increase from
liquid to solid is very slight, but not negligible (0.1%), that there is no dendritic growth during
solidification, and that the metal is pure.