CONGA: THE WORLD’S FIRST 42 FOOT DIAMETER 28 MW GEARLESS SAG MILL

T. Orser1, V. Svalbonas2 and *M. Van de Vijfeijken 3

1
Newmont Mining Corporation
10101 East Dry Creek Road
Englewood, CO 80112, USA
2
Metso Corp.
2715 Pleasant Valley Rd.
York, PA 17402, USA
3
ABB Switzerland Ltd
Segelhofstrasse 9 P
Baden-Daetwill, Switzerland, CH-5401
(*Corresponding author: [email protected])

1
 CONGA: THE WORLD’S FIRST 42 FOOT DIAMETER 28 MW GEARLESS SAG MILL

ABSTRACT

In 2010 Newmont made the decision to purchase a 42 foot (ft) Semi-Autogenous Grinding (SAG)
mill from Metso and a 28 Mega-Watt (MW) Gearless Mill Drive (GMD) from ABB for the Conga Project
in Peru. The test and study work showed that a single line 42 ft SAG mill circuit gave the most favorable
project economics in comparison to an equivalently powered single line 40 ft SAG mill circuit. This paper
does not address the process design or the project economics, but instead focuses on the key aspects of the
SAG mill design, GMD design and the associated component manufacturing. This paper also addresses,
where applicable, the potential reliability differences between the initially proposed 40 ft SAG mill and the
finally accepted 42 ft SAG mill.

KEYWORDS

SAG mill, 40 ft, 42 ft, gearless mill drive, GMD, SAG manufacture, reliability

THE PROJECT

The Conga Project is located in Peru adjacent to Newmont’s existing Yanacocha operations. The
Conga Project has been in various development stages for approximately the last 15 years. In the last 5
years, test and study works were performed to determine the optimal mill circuit configuration and mill
sizes. The results of these studies showed that a single line 40 ft SAG mill circuit gave the best project
economics. Further work was followed up to optimize the 40 ft SAG mill size with the results showing
that a slightly more expensive 42 ft SAG mill with equivalent power to the 40 ft SAG mill enhanced the
project economics. After assessing the potential reliability difference between the 40 ft SAG mill and the
equivalently powered 42 ft SAG mill, Newmont made the decision in 2010 to purchase a 42 ft SAG mill
from Metso and a 28 MW GMD from ABB.

THE MILL

Needless to say “reliability” is a key word at every stage of the project. Although mill structure
failures have not occurred in large GMD SAG mills, the mill design was not exempt from a reliability
survey. This survey, or assessment was carried out by Metso partly in conjunction with Newmont, and
included the following phases:

* A survey of engineering company mill specifications, looking mainly for obsolete areas;
* A survey of manufacturing facilities currently involved in 40 ft mills; and
* An overall survey of the design of the major mill rotating components.

A summary of the survey results is provided below.

Mill Specification Survey

In an overview of 7 engineering company specification documents, over a variety of projects over

the last 2.5 years, all specifications were found to be essentially “dead documents”, with any engineering
personnel interested in serious updating of the technology descriptions having long since retired. Some of
the obsolete items found are listed below:

2
i. “Do not use any process not previously used on mills”. Considering this statement is from
1980, no new fabrication (welding, etc.) technology has been allowed for 31 years. One may
wonder how the stainless clad mills were ever built under this specification statement.

ii. “Use only the single wire welding processes”. This statement comes from 1983. Tandem
wire procedures have long since been perfected, and save both time and money in mill
fabrication. They have been used on mills since 1990.

iii. The fracture initiation ( KTH) values, where given, come from 1986, and are drawn from an
old document (Speidel 1981) for a different class of iron, and inappropriate r (tension to
compression) ratio. Since then, both design companies, and foundries, have done testing
(specified by ASTM) to obtain more appropriate values, matching the iron classes.

iv. “Map all dross”, more recently co-opted to “no dross”. The wording for this arises from 1987,
when larger SGI castings started to be made, and were substituted for grey iron. The wording
came from experiences with the Kennecott 34 ft SAG mills, where large dross areas were
encountered on the head internal conical surfaces, for which no machining had been planned.
Since this was a relatively new flaw (to mining) in 1987, the specifications were re-written to
plan for stock allowances to remove dross in machining, and map what might remain. It is
useful to note that the Kennecott 34 ft mills (and a number of other large mills with
unmachined cones) continue to run today, after 24 years, justifying the reasoning to accept
those castings “as is”. Since that time, much testing has been done on dross, at the foundries,
and at Metso, so that this flaw is now well understood.

v. Inappropriate use of BS7608 (1993) fatigue limits combined with “peak stress” terminology.
BS7608 explicitly states, on p. 30, “Hence the hot spot stress is considerably lower than the
peak stress, but provides a consistent definition of stress range for application with the design
S-N curves.” This is most often misused in reference to ground weld surfaces. BS7608 is a
code for the design of various welded structures, not mills. In being an empirical code drawn
from testing, it does not analyze individual weld ripples and crevices, but rather includes
them in the test data by using as-welded specimen which satisfies visual criteria. The code
considers diversity in these geometries by using statistical S-N curves. As such, the code idea
of grinding weld surfaces requires no defined shapes or a minimum radius, etc., and thus
provides only a small (in stress terms) improvement to the hot spot allowable stress. For
further explanation, refer to Figures 1 and 2. If the smoothly ground weld in Figure 1 has just
one small shallow area, where, due to the grinder slippage, there is a slight 12mm radius
depression, the peak stress here could more than double. This is NOT what the code considers
in ground welded structures. The local peak stress here is analogous to the local dimple
stresses not considered in as welded conditions. The basic definition of “hot spot stress”
holds for ground welds also.

3
 Figure 1 – FEA of ground weld

Figure 2 – FEA of ground weld with overgrind

4
 vi. “Design…shall include a ball charge of a least 3%...greater than the maximum operating (sic)
ball charge…, but not less than 18% of the total volume.” This states that AG mills are
designed at 18% ball safety, mills operating at maximum 12% balls get 6% ball safety, while
operations at maximum 18% get 3% ball safety. The original statement was that all mills
should be designed at 3% ball safety over their maximum operating load.

Some of the obsolescences listed above overlap into the choice of consultants. There are almost
no consultants in the mill industry that actively pursue ‘hands on’ research in the materials and flaw
sensitivity areas. Yet the original formulations were developed in the 80’s. Since then, a large amount of
research has been performed in these areas, as they apply to mill constructions. This new state of the art
has to be considered. Churchill once stated:

“However beautiful the strategy, you should occasionally look at the results.”

This may be paraphrased for mill analysis:

“If your theory tells you all the 20+ year old successfully operating mills failed 10 months after
start up, you need a new consultant.”

Manufacturing Facilities Survey

In direct contrast to the unchanging areas of mill specifications, the survey of manufacturing
facilities found some interesting items which might be related to a change (relaxation) of surveillance. In
the days of the first gearless mills of the ‘80s, and thru the first 40 ft mills, extensive fabrication kick-off
meetings were held, where all processes were discussed. Perhaps because this has become familiar, such is
no longer the norm. Because of this, and the retirement of experienced personnel, some interesting
fabrication changes were discovered.

i. At one European foundry, head machining is subcontracted. In a review of these capabilities,

a change in the machining procedure was discovered. Whereas the usual head machining is
accomplished in one set-up, without moving the piece, in this case two machining set-ups
were being used. Where previously this required picking up the assembled, partly-machined
head, and turning it over as a unit, in this case, 40 ft heads were being completely
disassembled, and turned over segment by segment, and then being reassembled. No written
procedure of how the original machining cut inter-relationships were to be reestablished,
before commencing further cutting in the new assembly, were produced. Therefore, this
machining method was not accepted by Newmont and Metso for the Conga 42 ft mill. One of
the authors (VS) had experience with this kind of reassembly, for measuring purposes only,
and established tolerances in one position were not being duplicated in the reassembly, after
days of effort. In light of specification item (i), it is interesting that this change was missed
by inspectors.

ii. The lack of capability audits seems to originate with company buyers. The possibility of 20%
+ price reductions seems to overcome all quality and design audit requirements, even for
large mills. This has allowed such undesirable fabrication practices to enter the industry as
pictured in Figures 3 – 5. Figure 3 shows uncontrolled heat correction for a large shell.
Figure 4 shows “vibratory stress relief” being applied to a large shell. If this method is to
work, room temperature material yield stress must be exceeded for a large number of cycles,
which reduces the shell operating fatigue life by the same amount. Figure 5 shows significant
weld overlay being added to critical steel head flanges, to make up for errors in estimating
casting shrinkages. This is compounded by the fact that specifications have not incorporated
the detailed changes required to evaluate the flaw sensitivity of Chinese steel castings. The
experiences gained with ductile irons cannot be directly applied to these new materials.

5
Figure 3 – Uncontrolled heat straightening

Figure 4 – Vibration “stress relief”

6
 Figure 5 – Weld overlay on flanges

This cost pressure has triggered a cost cutting competition also in western mills, which has
resulted in both fabrication and design changes. In the latter area, previous increases in
allowable stresses were only pursued after detailed field experience. Currently, they are
being pushed into larger mills with much less study, as illustrated in the examples below:

Table 1 - Changing design stress

SAG Internal Weld Design
Mill Size Design Stress Age of Operation
Los Bronces 10.36m 58.5 MPa 23 years
Bata Hijau (sea water) 10.97m 48.2 MPa 14 years
Esperanza (sea water) 12.2m 69 MPa 6 months
Conga 12.8m 53 MPa -

Ball Mill External Weld Design

Cadia 6.7m 37.8 MPa 14 years
Conga 7.26m 38 MPa * -
Toromocho 7.82m 51.4 MPa * -
* Strain gage correlation has been applied

For the Conga mill design, experience defined stresses were used.

iii. The last several years have reintroduced plate welding problems. Previous plate ordering
specifications were no longer working to eliminate lamination and void problems. This
comes about from greatly increased use of the continuous casting process in the steel industry.
This process generally produces cheaper plate, but with poorer characteristics at greater
thicknesses. This is complicated by the fact that individual steel makers use different
thicknesses to separate the slab (continuous casting) from ingot making methods. For

7
example, for Arcelor Mittal (North America) this boundary thickness is about 112 mm,
whereas Dillinger may use continuous casting methods up to 160 mm thickness. This is
further complicated in that the reduction ratio for some of these slab processes may be
significantly lower than 3. Thus, for the Conga mill, the fabricators were warned of these
possibilities, and requested to have discussions with their plate suppliers. One of the
problems that can arise from this omission is shown in Figure 6. The figure shows part of a
contour flange, which passed all straight beam ultrasonic tests, but where center-line plate
defects were discovered by angle beam inspection of adjacent welds. Since this cannot be
eliminated by standard ultrasonic inspections accepted by steel mills, it must be eliminated by
a change in ordering specifications.

Figure 6 – Flaw location in contour flange

8
Major Mill Rotating Component Design Survey

The key driver with this survey was to design the 42 ft SAG mill heads and shells for maximum
reliability by using the existing proven 40 ft SAG mill design concepts. In performing the Conga detailed
design, several questions were asked:

1. How can we keep to the known design art as much as possible, minimizing “prototyping”?

2. Are there any recent design developments to be considered?

In answering the first question, the mill components were rated as to risk: large castings were
rated as higher risk than shell fabrication, and so forth. The challenge was then to make the components
have no greater risk than a 12.2m (40 ft) mill. This was accomplished in the following way:

The 12.8m (42 ft) mill head castings were designed to be of no more weight than 12.2m (40
ft) heads, and of no larger diameter;
The shell fabrication, while 0.6m (2 ft) larger in diameter, used plates of similar thickness as
found on 12.2m (40 ft) shells; and
The contour flange was designed deeper, but only of slightly thicker plate.

In this way, the mill construction “prototyping” was limited to manufacturing the shell at the
12.8m (42 ft) diameter. The machining of the components was accomplished on the same machines as for
12.2m (40 ft), and purchasing of parts was not affected, except to the level of control of plate purchasing
that was noted earlier.

In answering the second question, a survey of design details was done. Two interesting items
were uncovered:

(i) Segmentation of larger ball mills had produced a stress concentration point, associated with
welding, as shown on Figure 7. While this point had been modeled (finite elements), an
instrumentation program for validation had not been done. This was performed on two
mills during 2010. The strain gauges verified the existence of a significant stress multiplier
at the noted point. The strain gauging program also verified a known fact, that the standard
FEA model produced very conservative results in this area. However, combining both
results still showed that the noted stress concentration point could not be ignored.
Therefore, design variables were adjusted for mills of this segmentation.

Figure 7 – Stress concentration point

9
 (ii) In the general information search, a new proposal to the ASME code was uncovered. In
piping welds, tests had shown that reinforcing weld fillet overall profiles could have
significant fatigue life effects. It was shown that if the leg along the pipe is 2x that of the
flange leg, the fatigue life could improve significantly (up to 37%). Some consideration of
this was being proposed for the ASME code. Conveniently, Metso standard welding on
large mill flanges had always utilized such leg dimension bias, and thus extra safety was
verified for standard methods. For the mill, being a “large pipe”, items (i) and (ii) become
inter-related. Since item (ii) indicated extra conservancy (but in the early stages of study),
no numerical advantage of this was incorporated into the present designs.

THE GEARLESS MILL DRIVE

The gearless motor (also called wrap-around motor or ring motor) is a very large synchronous
motor. The poles of the motor are directly installed on a pole flange on the mill shell; this means the mill
body becomes the rotor. The stator of the gearless motor is then wrapped around the mill. Figure 8 shows
a picture of the 16.5 MW ABB gearless motor installed on the 32 ft diameter Metso SAG mill at
Newmont’s Yanacocha mine in Peru.

Figure 8 – Yanacocha gearless motor

The gearless motor is fed by a cyclo converter, which is an extremely compact and efficient drive
system, opening the door for several advantages such as variable speed and frozen charge protection. With
this innovative concept, the gearless mill drive (GMD) eliminates all critical mechanical components
inherent within a conventional mill drive system, such as ring gear, pinion, gearbox, coupling or air clutch,
motor shaft and motor bearings. Eliminating such components increases the efficiency and the availability
of the mill.

10
Brief History of Gearless Mill Drives

The world’s very first GMD was a 6.4 MW GMD delivered by ABB (BBC, before the merger
with Asea) for a cement mill. This GMD started operation in 1969 and is still in operation today after
more than 40 years. In the 1980s, the GMD was introduced to the mining industry as a reliable drive
system for SAG mills. Early SAG mills had a diameter of approximately 32 ft, later increasing to 34 ft, 36
ft, 38 ft and 40 ft and were quite reliable. In the last decade some issues occurred with the larger GMDs
which resulted in unscheduled downtimes. Subsequently, improved design and operational practices have
been implemented in the last several years. Despite these issues many large mills have been ordered with
some already commissioned and in operation. Currently, the largest GMDs in operation are the two 22.5
MW GMDs for the two 38 ft diameter x 45 ft length AG mills at Boliden’s Aitik copper concentrator
above the Arctic Circle in north Sweden and the 22.4 MW GMD for the 40 ft SAG mill of Esperanza in
Chile. Besides these large GMDs in operation, ABB last year delivered a 28 MW GMD for the 40 ft SAG
mill on the Toromocho Project in Peru, which will be installed next year at an elevation of 4,600 meters
above sea level (masl) or over 15,000 ft. Not only ABB, but also another GMD supplier has delivered a 28
MW GMD for a 40 ft mill for an iron ore project in Australia. There is not only a clear trend in the global
grinding industry for large diameter mills with high power GMDs, but also for mills requiring GMDs with
high power density. The GMD power density, as defined in this paper, is a qualitative measure calculated
by taking the mill rated power [MW] and dividing by the mill diameter [ft]:

mill rated power [MW ]

GMD power density =
mill diameter [ft ]

Figure 9 shows that the power density of the GMDs has continuously been increasing during the
past 15 years for both the SAG and ball mills.

0.8000

0.7500

0.7000

0.6500
Power density [MW/ft]

0.6000
(S)AG
0.5500
Ball

0.5000

0.4500

0.4000

0.3500

0.3000
1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
Order intake

Figure 9 – Power density of ABB GMDs in the mining industry

There are several reasons for this trend of increasing power density values. For example, mills are
getting longer and ball charges are getting higher. Most likely this is also driven by the fact that the ore

11
bodies today are having lower ore grades and the need to maximize production rates. The trend of larger
mills with higher power requirements and higher power density values naturally impacts the GMD design.

The GMD for the Conga Project

Since the step was made to go from 38 ft to 40 ft mills, the mining industry took more than a
decade to go to 42 ft mills. The Conga SAG mill drive requirements are as follows:

- Diameter: 42 ft
- Rated power: 28 MW
- Rated speed: 8.86 rpm (74% critical speed)
- Maximum speed: 10.0 rpm
- Rated torque: 30,178 kNm
- Starting torque: 150%

The power density of the Conga 28 MW GMD is 0.67. Figure 9 shows that this power density is
not the largest value or an industry first. In addition, the Conga concentrator will be located at a relatively
high elevation of 4,100 masl, but again not the highest elevation or an industry first. All these important
parameters (the 42 ft diameter, the 28 MW rated power, the power density, and the site elevation) have
certain impacts on the GMD design and overall reliability.

42 ft Diameter

As already mentioned, some relatively large GMDs had problems for various reasons. The main
goals for this first 42 ft diameter GMD are that no known past issues will resurface again and to design
where the potential for new issues are eliminated. The starting design reference is based upon the 28 MW
GMD for the Toromocho 40 ft SAG mill for which several design review have been performed.

It has to be noted that with increasing diameter, the first possible resonance frequency (4-node
bending mode) is decreasing if no counter measures are taken. The correct design philosophy is obviously
to stay below the first possible resonance frequency, especially considering the fact that the stator structure
has a height of more than 20 meters. In order to define the resonance frequencies, a finite element model
of the stator is created. The risk is the finite element model is for some reason not reflecting reality and
consequently the calculated resonance frequencies could be wrong. In the last decade, the finite element
model has been verified with a tapping test in the workshop on almost all GMDs delivered by ABB. On
some occasions the tapping test has even been performed on site. On the 28 MW GMD for the 40 ft SAG
mill for Toromocho, the finite element model was verified with a tapping test in the factory. On some
GMDs site strain gauge measurements have been performed also to verify the model. Tapping tests in the
factory and field will be performed on the Conga GMD along with site strain gauge testing. ABB has also
performed the overall system study on several projects. This study analyzes the overall stiffness and
resonance components of the soil, foundations, mill bearings, mill, rotor, magnetic pull between stator and
rotor, and the stator. ABB will perform the system study on the Conga SAG mill. A 3 rd party system study
will also be performed to verify ABB’s system study. The past experiences with the tapping tests, strain
gauge tests, and overall system studies have been incorporated into the design considerations of the Conga
SAG mill GMD.

28 MW Rated Power

Before the Conga Project, ABB received two other orders for a 28 MW GMD, but with a smaller
mill diameter of 40 ft: Toromocho in Peru and Boszhakol in Kazakhstan, which both have been
manufactured and delivered. So the Conga GMD has exactly the same power rating as for the Toromocho
and Boszhakol Project. Naturally this high power requirement has several impacts on the design. The
standard ABB GMD design is typically realized with a 12-pulse cyclo converter, but this is only available
up to approximately 26 MW rated power. In order to reach more than 26 MW, either the stator voltage or

12
the stator current (or both) have to be increased. Several years ago, ABB performed the basic design for
high power GMDs up to 36 MW. ABB decided then to stay with the same proven electrical concept of the
gearless motor. In order to reach the 28 MW, only an extra leg with thyristors had to be added resulting in
an 18-pulse drive system as shown in Figure 10.

Figure 10 – Block diagram of the 18-pulse GMD

The advantages are as follows: available true motor differential protection, relatively simple
winding design, minimized number of thyristors, just one cyclo converter instead of two cyclo converters
which would require a kind of load sharing and reduced total harmonic distortion. Although the Conga 28
MW drive system is large, the basic design concept is proven and is being used on other projects.

Power Density

The 42 ft diameter Conga SAG mill GMD power density value of 0.67 is not an industry first or a
new record. The 40 ft diameter Toromocho SAG mill GMD has a higher power density value of 0.70 at
the same rated power of 28 MW. The 22 MW GMD for the 28 ft diameter Toromocho ball mill has an
even higher power density value of 0.79. The power density has quite an impact on the design. More
power means more active material in the motor. This means the motor size must increase, but unlike
conventional electric motors the motor cannot be increased in radial direction (“increasing shaft height”).
Basically, the gearless motor can only increase in the axial direction. In other words, a higher power
density value results into a wider motor (in axial direction). For the stator this means that both the stator
windings (due to the iron length) and the magnetic core lamination package are longer. For the rotor this

13
means the pole units are getting longer. The poles will be single pole units without critical welding where
every single pole can be adjusted slightly in order to have a perfect run out. The nominal air gap around
the circumference is about 17 mm and an imperfect run out could lead to undesirable oscillating mill
bearing pressures.

A wider motor can also have consequences on the mill design, especially on the last row of liner
bolts as access to these bolts must still be ensured in order to be able to replace the liners. This challenge is
visualized in Figure 11 below and requires close cooperation during the design phase between the mill
supplier and the GMD supplier. Often the last row of liner bolts are mounted under a certain angle.

Figure 11– Challenge to access the last row of liner bolts

In addition to the challenge of removing the last row of liner bolts, a wider motor also challenges
the conventional cooling design concept of the gearless motor. Typically the gearless motor has an internal
axial cooling system where the cooling air flows in axial direction through the lamination package. This
means there are certain limitations to the maximal length of the lamination package because otherwise the
cold air to cool the motor will already have become too warm before it reaches the other side. Therefore a
radial cooling concept is used for all the gearless motors with a high power density value: in this case the
cooling air flows in radial direction through the lamination package. Exactly the same cooling concept is
used on other large synchronous machines such as low speed diesel generators and hydro generators. It has
to be noted that apart from this modification from axial to radial cooling, the cooling principle of the GMD
has not changed. The cooling boxes with the water-to-air heat exchangers are still located at the bottom
below the stator windings which means there is no cooling water at all around the circumference. If a leak
occurs water cannot reach the stator windings and cause a serious problem. The Conga GMD will use the
proven radial cooling design concept.

Site Elevation

The Conga mine site is located at an elevation of 4,100 masl, and this naturally places additional
demands on the design. The high altitude has an impact on the efficiency of the cooling air. Furthermore,
if not handled correctly, the high altitude could have an accelerated ageing effect on the electrical
insulation. All these important parameters have been considered in the Conga GMD design. In addition,
the extensive experiences gained with GMDs at altitudes over 4,000 masl have also been used in the design.
The 28 MW GMD for the 40 ft SAG mill of the Toromocho Project has been delivered and will be
installed at an even higher altitude of 4,600 masl (over 15,000 ft) in the Peruvian Andes. Approximately
two years ago, several high altitude insulation tests of a stator part complete with windings were conducted
in Switzerland in a hypobaric chamber, simulating conditions found at 5,000 masl. These tests were also
witnessed by several customers. The test setup is shown in Figure 12. All tests were satisfactory passed.

14
 Figure 12 – Insulation tests in hypobaric chamber simulated at 5,000 masl

CONCLUSIONS

Newmont, Metso and ABB have given special attention to the important fact that the Conga SAG
mill will be the first 42 ft diameter SAG mill in the world. Reliability assessments were done concerning
mill specifications, manufacturing facilities, critical component designs, and power limitations. The
assessments bare out that the Conga SAG mill will be an industry first in diameter, but not necessarily an
industry first in rated power, overall design, and component manufacturing. The mill and GMD are
scheduled to be delivered in the middle of 2012 and go into operation in late 2014, early 2015. The
successful operation of the world’s first 42 ft SAG mill will be the culmination of a continuing tradition in
the mining industry of building larger equipment to enhance project economics.

REFERENCES

Speidel, M.D., (1981), Z. Werkstofftech, vol. 12, pp. 387-402.

BS7608, (1993), “Code of Practice for Fatigue Design and Assessment of Steel Structures”, British
Standard.

Private communications (2010), Arcelor Mittal North America.

Private communications (2011), Metso, ALC, Dillinger Hutte.

15