DIRECT VS.

REVERSE GRINDING CIRCUIT

CONFIGURATIONS: A QUANTITATIVE ASSESSMENT OF
THEIR RELATIVE OPERATIONAL PERFORMANCE
Jaime Seplveda
Moly-Cop Grinding Systems / SCAW Metals Group, Chile

ABSTRACT
Even though the classical DIRECT or REVERSE configurations of industrial grinding/
classification circuits must be as old as ball milling technology itself, the uncertainty
as to which circuit configuration would be more effective, for a given set of grinding
conditions, still arises recurrently amongst process designers and mill operators. There
exists a generalized perception that the classical REVERSE configuration is intrinsically
more productive than the alternative DIRECT configuration. The purpose of this publication
is to provide a simple and quantitative, discriminating criterion to resolve this query, in
the context of the so-called Fourth Law of Grinding/Classification; which states that in
order to obtain optimal energy efficiency in the overall grinding process, the content of
fine particles in the mill charge must be as low as possible, for a given grinding task. As
a rule of thumb, it is here claimed that the REVERSE configuration will be advantageous
only when the fresh feed stream contains more than 30% of mineral particles finer than
the target P80 size specified for the operation. In practice, such critical condition is
only achieved in quite few special cases, like secondary mills after rod mills or ball mills
following true SAG mills; cases in which the particles fed to the mill would be fine
enough. It is so concluded that the DIRECT circuit configuration should be the preferred
arrangement in most practical cases, unless other circuit layout considerations prevail.

08 PRM_Book PDF.indb 55

14-10-2008 9:32:59

CHAPTER 02

The Fourth Law Optimality Claim

Even though the classical DIRECT or REVERSE configurations of industrial grinding/
classification circuits (see Figures 1a and 1b) must be as old as ball milling technology itself, the uncertainty as to which circuit configuration would be more effective, for a given
set of grinding conditions, still arises recurrently amongst process designers and mill operators. There exists a generalized perception that the REVERSE configuration shown in
Figure 1b (where the fresh feed slurry is charged to the pump sump) is intrinsically more
productive than the DIRECT configuration shown in Figure 1a (where the fresh feed slurry is charged directly to the mill). At least at first look, it seems reasonable to believe that
if some fraction of the feed ore corresponds to fine enough, product size particles, it would
be effective to allow them to by-pass the mill by feeding them to the cyclones first. But,
does it always make sense? Under what specific operating conditions could one configuration be advantageous over the other?
The purpose of the present technical note is to provide a simple and quantitative,
discriminating criterion to resolve these concerns, in the context of the so called Fourth
Lawof Grinding/Classification [1, 2, 3]; which states that in order to obtain optimal energy
efficiency in the overall grinding process, the content of fine particles in the mill charge must
be as low as possible, for a given grinding task. Achieving this condition of reduced fines
content in the mill charge allows to spend most of the available energy in grinding the
predominant coarser particles and at the same time, avoid the overgrinding of the nonpredominant finer particles. On this basis, it is claimed here that the preferred circuit
configuration is the one that would maintain the mill with the lowest content of fine
particles in its charge.

Figure 1a: Typical direct grinding/classification circuit configuration

Figure 1b: Typical reverse grinding/classification circuit configuration

56

08 PRM_Book PDF.indb 56

14-10-2008 9:33:00

Direct vs. Reverse Grinding Circuit Configurations: A Quantitative...

The Fourth Law Mass Balance Equations

With reference to Figure 1a, where MSk represents the dry massflowrate (tons/hr) of
mineral particles flowing through stream k, at steady state conditions, the following mass
balance relations must apply:
MS1 + MS6 = MS2
MS2 = MS3
MS3 = MS5
MS5 = MS6 + MS7
CL = MS6 / MS7

(1)

The last equation is just the definition of the circuit Circulating Load (CL). The above
system of 5 equations can be readily solved in terms of MS1 and CL, as independent variables, to obtain:
MS1 = MS1
MS2 = (1+CL) MS1
MS3 = (1+CL) MS1
MS5 = (1+CL) MS1
MS6 = (CL) MS1
MS7 = MS1

(2)

Then, focusing the attention on the finest particles in every stream (say, the -400# or -325#
size fractions), with mk representing the mass flowrate (ton/hr) of such fines in every
stream k, at steady state conditions, the following mass balance relations must also apply:
m1 + m6 = m2
m3 = m5
m5 = m6 + m7
Bpf = m6 /m5

(3)

with Bpf being the definition of the cyclone solids by-pass ratio; that is, the fraction of the
cyclone feed slurry that short-circuits directly down to the cyclone underflow.
The last three groups of equations above can be solved to obtain a direct relationship between the fines flowrate in the cyclone overflow (m7) and the mill discharge (m3) streams:
m3 = m7 / (1 Bpf)

(4)

and from where, dividing each side of this equation by MS3 = (1+CL) MS1, the following
expression is finally obtained:
F3 = F7 / (1-Bpf) / (1+CL)

(5)

with F3 = m3 /MS3 and F7 = m7/MS7, representing the fractional content of fines in the
mill discharge and cyclone overflow streams, respectively.
Therefore, referring back to the Fourth Law criterion, a DIRECT circuit configuration
will operate at improved overall energy efficiencies as the short-circuit ratio (Bpf) decreases and the Circulating Load (CL) increases as, per Equation 5, these trends should decrease the content of fines in the mill discharge stream (F3); the best available indication
of the fines content in the load of mineral particles inside the mill.
Following a completely analogous sequence of steps, this time for the REVERSE
PROCEMIN 2008. Santiago, Chile

08 PRM_Book PDF.indb 57

57

14-10-2008 9:33:00

CHAPTER 02

configuration (Figure 1b), an equivalent set of mass balance equations are also
applicable:
MS1 + MS3 = MS5
MS5 = MS6 + MS7
MS6 = MS2
MS2 = MS3
CL = MS6/MS7

(6)

which can be similarly solved, in terms of MS1 and CL, to obtain:

MS1 = MS1
MS2 = (CL) MS1
MS3 = (CL) MS1
MS5 = (1+CL) MS1
MS6 = (CL) MS1
MS7 = MS1
And also, for the finest particles in
should apply:
m1
m5
m2
Bpf

(7)

the system, the following mass balance relations

+
=
=
=

m3 = m5
m6 + m7
m6
m6/m5

(8)

Equations 8 above can be solved to obtain a direct relationship between the fines flowrate in the cyclone overflow (m7), the fresh circuit feed (m1) and the mill discharge (m3)
streams:
m3 = m7 / (1 Bpf ) m1

(9)

from where, dividing each side of this equation by MS3 = (CL) MS1, the following expression for F3 can be derived for the REVERSE circuit configuration:
F3 = F7 / (1-Bpf ) / CL - F1 / CL

(10)

with F1 representing the fraction of fines in the fresh feed ore; a term that is not present
in the previous comparable Equation 5.
Similar to a DIRECT arrangement, a REVERSE circuit configuration will operate at improved overall energy efficiencies as the short-circuit ratio (Bpf ) decreases and the Circulating Load (CL) increases as, per Equation 10, these trends would decrease the content of
fines in the mill discharge stream (F3).

Optimality Criterion
Finally, in the context of the Fourth Law criterion that is, in order to obtain optimal energy
efficiency in the overall grinding process, the content of fine particles in the mill charge must be
as low as possible, for a given grinding task the REVERSE circuit configuration will be the
preferred configuration whenever:
F3 REVERSE < F3 DIRECT

(11)

58

08 PRM_Book PDF.indb 58

14-10-2008 9:33:00

Direct vs. Reverse Grinding Circuit Configurations: A Quantitative...

From the previous Equations 5 and 10 assuming both configurations would operate at
similar Bpf and CL values this leads to the critical discriminating condition:
F7 / (1-Bpf ) / CL - F1 / CL < F7 / (1-Bpf ) / (1+CL)

(12)

which may be rearranged as:

F1 > F7 /(1-B pf ) / (1+CL)

(13)

Figure 2 was constructed on the basis of Equation 13 to illustrate under what conditions
the REVERSE configuration would be preferred (zone above the corresponding curve)
and the DIRECT configuration would be preferred (zone below the corresponding curve).
For typical values of Bpf = 0.4 (40% Cyclones By-pass) and CL = 3.5 (350% Circulating
Load), it is concluded that:
F1 > 0.37 F7

(14)

that is, the fraction of fines in the fresh feed ore (F1) must be at least 37% of the fraction
of fines contained in the final ground product overflow stream (F7), for the REVERSE
circuit to be the preferred circuit configuration option.

Figure 2: Critical F1 /F7 ratio as a function of the cyclones short-circuit and circulating
load ratios. Above the curves, the REVERSE configuration should be preferred

As an approximation, Equation 14 can also be expressed in terms of the P80 size of the
final ground product assuming that the weight ratio of the % passing 400# or % passing
325# in the fresh feed (stream 1) and the cyclone overflow (stream 7) is similar enough
to the weight ratio of material passing the P80 size in the same two streams, which can be
expressed as:
F1 (fines) / F7 (fines) F1(P80) / F7 (P80)

(15)

and since, by definition F7 (P80) = 0.80, then Equation 15 becomes:

F1 (P80 ) > 0.8 /(1-Bpf ) / (1+CL)

(16)

which, for the same typical values of Bpf = 0.4 and CL = 3.5, yields:
F1(P80) > 30%

PROCEMIN 2008. Santiago, Chile

08 PRM_Book PDF.indb 59

(17)

59

14-10-2008 9:33:00

CHAPTER 02

Final Remark
In actual practice, the critical condition in favor of the REVERSE configuration (Equation
13) can only be achieved in quite few special cases, like secondary mills after rod mills or
ball mills following true SAG mills (i. e., with less than 10-12% balls and no pebble ports
in the their grates); cases in which the particles fed to the mill would be fine enough. It is
so concluded that the DIRECT circuit configuration should be the preferred configuration
in most practical cases, unless other circuit layout considerations prevail.

REFERENCES
Seplveda, J. (1984). Una Cuarta Ley para la Molienda /clasificacin. Proceedings of the IV
ARMCO Mineral Processing Symposium, Via del Mar, Chile. [1]
Seplveda, J. (1990). Ball Mill Grinding: 40 Years after Bond. Proceedings of the VI ARMCO
Mineral Processing Symposium, Via del Mar, Chile. [2]
Gutirrez, L. & Seplveda, J. (1986). Dimensionamiento y optimizacin de plantas concentradoras
mediante tcnicas de modelacin matemtica. CIMM-Chile Publication. [3]

60

08 PRM_Book PDF.indb 60

14-10-2008 9:33:01