附件1：外文资料翻译译文

非圆柱滚子破碎机的设计

摘要
   低的破碎比和高的磨损率是与传统的破碎机相联系的很常见的两个特性。因为这点，在矿石处理流程的应用中，很少考虑到它们，并且忽略了很多它们的优点。本文描述了一个已被发展起来的新颖的对辊破碎机，旨在提出这些论点。作为NCRC,这种新式破碎机结合了两个辊筒，它们由一个交替布置的平面和一个凸的或者凹的表面组成。这种独特的辊筒外形提高了啮合角，使NCRC可以达到比传统辊式破碎机更高的破碎比。用一个模型样机做的试验表明：即使对于非常硬的矿石，破碎比任可以超过10。另外，既然在NCRC的破碎处理中结合了辊式和颚式破碎机的作用，那就有一种可能：那种新的轮廓会带来辊子磨损率的降低。

介绍
传统的辊筒破碎机因为具有几个缺陷而导致了其在矿石处理应用中的不受欢迎。尤其是当与其它的一些破碎机比起来，诸如圆锥破碎机等，它们的低破碎比（一般局限在3以内）和高的磨损率使它们没有吸引力。然而，从矿石处理这一点来说，辊筒破碎机有一些非常可取的特点：辊筒破碎机的相对稳定的操作宽度可以很好控制产物粒度。弹簧承重的辊子的使用使这些机器容许不可破碎的物料（诸如夹杂金属等）。另外，辊筒破碎机是这样工作的：将物料牵引至辊子之间的挤压区而不是象圆锥和颚式破碎机那样依靠重力。这产生了一个连续的破碎周期，避免了高通过率，同时也使破碎机可处理潮湿的和胶粘的物料。
NCRC是一种新颖的破碎机，发明于澳大利亚西部大学，为得是提出一些与传统辊筒破碎机相联系的一些问题。新的破碎机结合了两个辊子，由间隔布置的平面和凸的或者凹的表面组成。这种独特的辊子轮廓提高了啮合角，使NCRC可达到比传统辊筒破碎机更高的破碎比。用一个模型样机的初步试验已表明：即使非常硬的物料，超过10的破碎比也可以实现。这些初期的发现是通过单一颗粒进给而获得的，在破碎中没有显著的物块间的相互作用。目前的工作在NCRC中用多物块试验延伸了现存的结果。同时也顾及了各种其他因素：影响NCRC特性和探索NCRC在选矿处理中使用效率。
操作原理
啮合角是影响辊筒破碎机性能的重要因素之一。小的啮合角是有利的，因为它们增大了物块被辊筒抓住的可能性。对于一个给定的入料粒度和辊隙，传统的辊筒破碎机的啮合角受限于辊筒的尺寸。NCRC试图通过有特殊轮廓的辊筒克服这种限制，这种轮廓提高了辊筒在一转中变化点的啮合角。至于啮合角，在选择辊面时，很多其他的因素，包括变化的辊隙，破碎的方式都考虑了。最终NCRC辊筒形状如图1所示。其中一个辊子由间隔布置的平面和凸面组成，而另一个是由间隔布置的平面和凹面组成。
NCRC辊筒的形状导致了几个独特的特点。其中最重要的就是在辊筒转动时，对于一个给定物块粒度和辊隙，NCRC所产生的啮合角将不再保持稳定。时而啮合角比相同尺寸的圆柱辊筒低很多，时而高很多。辊子转动中啮合角的实际变化量超过60度，如图2所示，图2也表示了相同情况下，可相比尺寸的圆柱辊筒破碎机所产生的啮合角。这些啮合角是对一个直径为25毫米的圆形物块放在辊径大约200毫米、最小辊隙1毫米的辊筒间计算出来的。这个例子可以用来描述使用非圆柱辊筒的潜在优点。为了抓住物块，通常啮合角不超过25度。因此，圆柱辊筒破碎机将一直夹不住这个物块，因为其实际啮合角一直稳定在52度。然而，在辊筒转过60度时，NCRC的啮合角降至25度以下。这意味着辊筒每转过一转，非圆柱辊筒破碎机可能有6次夹住物块。
试验过程
NCRC的实验室模型由两个辊筒部件组成，每一个由发动机、齿轮箱和有形辊筒组成。两个部件都安置在线性轴承上，其有效支持任何垂直部件的力，同时保证其水平运动。一个辊筒部件水平固定，而另一个通过压缩弹簧限制，压缩弹簧使辊筒抵抗一个变化的水平载荷。
可动辊筒上的预载荷可被调整直至最大值20千牛。驱动辊筒的两个电动机通过一个变化的速度控制器实现电同步，速度控制器使辊速连续变化直至14转每秒（大概0.14米每秒的线速度）。辊筒有一个188毫米的中心距，100毫米宽。两个驱动轴都装有应变规，用以测量辊筒扭矩。附加的传感器用以测量固定辊筒的水平力和辊隙。NCRC的边上装有透明玻璃以便于在运行是观察破碎区域，同时也使破碎流程得以用数码相机进行纪录。
试验进行于几种岩石，包括花岗岩、闪长岩、矿石、采石场弃石和混凝土。花岗岩和混凝土各取自商业性的采石场，前者先破碎、成形，而后者是爆炸的岩石。第一种矿石样品是SAG采石场进料，取于诺曼底煤矿的GGO，采石场弃石取于KAGMM煤矿。采石场弃石含有直径直至18毫米的金属颗粒，它们来自于经反复磨削和破碎的介质。混凝土由圆柱体样品（直径25毫米、高25毫米）组成，它们根据澳大利亚的有关标准制备。不受限制的单轴压力测试进行于矿山样本（直径25毫米、高25毫米），取于大量的矿石。结果表明：对于制备混凝土的强范围从60兆帕直至GG矿石样品的260兆帕。
起初，所有的样品都通过一个37.5毫米的过滤器去处任何粒度过大的物块。低于粒度要求的矿石被取样，并且过滤以决定入料粒度分布。在NCRC中每一个试验大约破碎2500克样品。这种样品粒度基于统计测试进行选择，那些统计测试表明： 为了估计百分之八十的通过率在正负0.1毫米范围内的百分之九十五的可靠度至少需要破碎2000克样品。选择并振动产品使其10次掉于过滤器下，使用一个标准的干的或湿的过滤方法以决定产品粒度分布。对于每一次试验，子样品中的两个被最先滤掉。如果产品粒度有任何显著的不同，额外的子样品将被滤掉。
使用NCRC进行大量的破碎试验以决定各种变化的参数的效果，参数包括：辊隙、辊上作用力、入料粒度和单个或多个物料进给。因为前面的试验以得出辊速对产品粒度分布影响很小，所以将辊速设定在最大值且前面两个试验之间不变。应该指出的是：辊隙设置引用提及的最小辊隙。因为辊筒的非圆柱体形，实际辊隙在设置的最小值以上的1.7毫米范围内变化（例：一个1毫米的辊隙设置值其意味着辊隙为1-2.7毫米）。

结果
入料
所有破碎设备的性能都依赖破碎物料的种类。在这方面，NCRC没有什么不同。在NCRC中破碎较软物料可产生低于较硬物料p80的碎强。图4所示是在NCRC中在相似条件下破碎几种不同物料时得到的产物粒度分布。有趣的是，除了备制混凝土样品外，从各种不同的物料中，p80碎强的获得也相当一致。结果反映：利用NCRC可获得对产物粒度分布的控制程度。
多入料物块
前面在NCRC上做的试验仅使用单入料物块，很少或没有物块间的相互作用。虽然很有效，但与这种破碎方式相联系的低的通过率不适合于实际应用。因此，决定连续进给对最终产品粒度分布的影响是有必要的。在这些测试中，连续供应以保持足够的物料以达到辊顶。图5显示，连续进给NCRC对诺曼底矿石产物粒度分布的影响。这些结果好像表明了使用连续（多物块）进给在p80碎强上的一个轻微的增加，然而变化太小以致其没有统计学意义。相似地，对于连续进给试验，产物粒度分布表明了一个较好结果，但实际上区别是微不足道的。如图6所示，用花岗岩样品使用不同的两个辊隙进行了相似的试验。又一次，在单个和多个物块测试间无变化。毫不夸张地，更大的辊隙、更小的破碎程度（物料间的相互作用），区别将更不明显。
所有的这些测试好像表明连续进给对NCRC的性能影响极小。然而，意识到在这些试验中用的进给物料在很小的范围内波动是重要的，如图6（诺曼底试验的进给物块甚至更一致）所示进给物块粒度分布。进给物块粒度的一致性导致了大量的自由空间，允许破碎腔内破碎矿石的增多，因此限制了物块间的相互作用。有一宽广物块粒度分布（尤其是较小的粒度范围）的带矿石的NCRC的真的“卡死”进给可能在破碎区域产生大得多的压力。既然NCRC不是作为“高压力破碎辊”而设计的，在这些情况下，更多的过大物块将从两辊间通过。
辊隙
象传统的辊筒破碎机一样，NCRC的辊隙设置对产品粒度分布和破碎机通过率有直接影响。图7展示了以三种不同辊隙破碎AG矿石（废弃矿石）时的最终产物粒度分布。针对辊隙从这张图中标出80值产生一线性关系，如图p8所示。如前解释所述，NCRC的实际辊隙将随着一转而变化。这一变化补偿了具体的辊隙设置和取于破碎试验中的产物百分之八十通过率间的差别。图8显示了辊隙对破碎机通过率的影响并给出了用NCRC的试验模型得到的破碎率。
辊动力
NCRC是利用煤块间的相互作用实现破碎机而设计的，这种破碎主要是通过直接折断辊间物块。因此，辊动力仅需足够大以克服辊面间物块的复合力。如果辊动力不够大，那么矿石块将分开辊筒，从而过粒度物块将落下。增大辊动力以减小辊筒分离倾向以更好控制产物粒度。然而，一旦达到限制辊动力（决定于被破碎物料的粒度和种类），辊动力的任何进一步增加都不能提高辊筒破碎机的性能。这由图9可得证，显示了25-31毫米的花岗岩入料，大约16-18千牛的辊动力去控制产物粒度。如果辊动力降至低于这一水平，虽然p80产物有一瞬间的增加，使用更大的辊动力对产物粒度仅有很小影响。
入料粒度分布
和前面提及的一样，入料粒度分布对破碎腔内产生的压力有明显影响。有更细的入料粒度分布的矿石更趋向于“卡死” NCRC，降低破碎机的效率。然而，只要所产生的压力不超过NCRC，不考虑入料粒度维持在一个相对稳定的操作间隙。因此，产物粒度分布也将不依赖于入料粒度分布。如图10所描述的，显示了使用相同的设备但不同的粒度分布的入料的两个破碎机试验的结果。在这个例子中，NCRC将较粗糙的矿石从80的34毫米破碎至80的3.0毫米（破碎比11：1），同时较细的矿石从80的18毫米破碎至80的3.4毫米（破碎比5：1）。这些结果表明，使用有形辊筒的缺点减少，同时，入料粒度和辊筒尺寸的比例在减小。另一方面，为了达到较高的破碎比，入料块度必须足够大以利用NCRC产生高的啮合角的优点。
废弃矿石
一些磨矿流程使用往复或石子破碎机（例如圆锥破碎机）去处理那些取自于选矿厂和发现难于破碎（废弃矿石）的物料。废弃矿石常含有坏的或破碎的磨粒，常见于往复破碎机中。因此，对于一个石子破碎机，不可破碎的公差是一个有意义的特性。NCRC看上去完美地适合于这一应用，既然其中一个辊筒能产生屈服以让不可破碎的物料通过。
图11所示的产物粒度分布取自于NCRC处理废弃矿石。对两个结果都使用相同的设备和入料粒度，然而，使用去处磨粒的矿石进行其中一个试验。和预料的一样，NCRC可以处理含有未进经INCIDENT的入料矿石。然而，既然一个辊筒为了让磨粒通过而经常移动，大量的未经破碎的过粒度物块可以通过辊隙。结果，这种入料粒度的产物粒度分布显示：对于更大的物块粒度的变化和P80值从4毫米增至4.7毫米。尽管如此，NCRC仍可以达到差不多4：1的破碎比。
磨损
    虽然没有对NCRC做具体的测试以决定磨损率，但为了试着了解破碎机理用高速录像机纪录了很多破碎试验。通过观察辊筒间被破碎物块，辊筒的部分区域好像受高磨损，并且得出一些主观结论：这种磨损对NCRC的性能有影响，这些都是可能的。毫不夸张地，所显示的高磨损的首要区域是平的和凹的过渡表面。令人惊讶的是，这种边缘在产生提高的啮合角方面不起重要作用。NCRC的性能不应该直接受这边磨损的影响，因为它实际上是平的和凸的表面的过渡区域（在辊筒的对面），导致了减小的啮合角。

附件2：外文原文
COMMINUTION IN A NON-CYLINDRICAL ROLL CRUSHER*
P. VELLETRI ~ and D.M. WEEDON ~
~[ Dept. of Mechanical & Materials Engineering, University of Western Australia, 35 Stirling Hwv,
Crawley 6009, Australia. E-mail piero@mech.uwa.edu.au
§ Faculty of Engineering and Physical Systems, Central Queensland University, PO Box 1!:;19,
Gladstone, Qld. 4680, Australia
(Received 3 May 2001; accepted 4 September 2001)
ABSTRACT
Low reduction ratios and high wear rates are the two characteristics ntost commonh" associated with conventional roll crushers. Because of this, roll crushers are not often considered Jor use in mineral processing circuits, attd many of their advantages are being largely overlooked. This paper describes a novel roll crusher that has been developed ipt order to address these issues.Relbrred to as the NCRC (Non-Cylindrical Roll Crusher), the new crusher incorporates two rolls comprised qf an alternating arrangement of platte attd convex or concave su@wes. These unique roll prqfiles improve the angle qf nip, enabling the NCRC to achieve higher reduction ratios than conventional roll crushers. Tests with a model prototype have indicated thar evell fi)r very hard ores, reduction ratios exceeding lO:l can be attained. In addition, since the comminution process in the NCRC combines the actions of roll arM jaw crushers there is a possibili O' that the new profiles may lead to reduced roll wear rates. © 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Comminution; crushing
INTRODUCTION
Conventional roll crushers suffer from several disadvantages that have lcd to their lack of popularity in mineral processing applications. In particular, their low reduction ratios (typically limited to about 3:1) and high wear rates make them unattractive when compared to other types of comminution equipment, such as
cone crushers. There are, however, some characteristics of roll crushers that are very desirable from a mineral processing point of view. The relatively constant operating gap in a roll crusher gives good control over product size. The use of spring-loaded rolls make these machines tolerant to uncrushable material (such as tramp metal). In addition, roll crushers work by drawing material into the compression region between the rolls and do not rely on gravitational feeci ~like cone and jaw crushers. This generates a continuous crushing cycle, which yields high throughput rates and also makes the crusher capable of processing wet and sticky ore. The NCRC is a novel roll crusher that has been dcveloped at the University of Western Australia in ordcr to address some of the problems associated with conventional roll crushers. The new crusher incorporates two
rolls comprised of an alternating arrangement of plane and convex or concave surfaccs. Thcse unique roll profiles improve the angle of nip, enabling the NCRC to achieve higher reduction ratios than conventional roll crushers. Preliminary tests with a model prototype have indicated that, even for very hard oics,
reduction ratios exceeding 10:I can be attained (Vellelri and Weedon, 2000). These initial findings were obtained for single particle feed. where there is no significant interaction between particles during comminution. The current work extends the existing results bv examining inulti-particle comminution inthe NCRC. It also looks at various othcr factors that influencc the perli~rmance of the NCRC and explores
the effectiveness of using the NCRC for the processing of mill scats.
PRINCIPLE OF OPERATION
The angle of nip is one of the main lectors effccting the performance of a roll crusher. Smaller nip angles
are beneficial since they increase tl~e likelihood of parlictes bcing grabbed and crushed by lhe rolls. For a
given feed size and roll gap, the nip angle in a conventional rtHl crusher is limited by the size of thc rolls.
The NCRC attempts to overcome this limitation through the use of profiled rolls, which improve the angle
of nip at various points during one cycle (or revolution) of the rolls. In addition to the nip angle, a number
of other factors including variation m roll gap and mode of commmution were considered when selecting
Ille roll profiles. The final shapes of the NCRC rolls are shown in Figure I. One of the rolls consists {sI an
alternating arrangement of plane and convex surfaces, while the other is formed from an alternating
arrangement of phme and concave surlaccs.

The shape of the rolls on the NCRC result in several unique characteristics. Tile most important is that, lk)r
a given particle size and roll gap, the nip angle generated m the NCRC will not remain constant as the rolls
rotate. There will be times when the nip angle is much lower than it would be for the same sized cylindrical
rolls and times when it will be much highcr. The actual variation in nip angle over a 60 degree roll rotation
is illustrated in Figure 2, which also shows the nip angle generated under similar conditions m a cylindrical
roll crusher of comparable size. These nip angles were calculated for a 25ram diameter circular particle
between roll of approximately 200ram diameter set at a I mm minimum gap. This example can be used to
illustrate the potential advantage of using non-cylindrical rolls. In order for a particle to be gripped, thc
angle of nip should normally not exceed 25 ° . Thus, the cylindrical roll crusher would never nip this
particle, since the actual nip angle remains constant at approximately 52 °. The nip angle generated by the
NCRC, however, tidls below 25 ° once as the rolls rotate by (~0 degrees. This means that the non-cylindrical
rolls have a possibility of nipping the particlc 6 times during one roll rewHution.
EXPERIMENTAL PROCEDURE
The laboratory scale prototype of the NCRC (Figure 3) consists of two roll units, each comprising a motor,
gearbox and profiled roll. Both units are mounted on linear bearings, which effectively support any vertical
componcnt of force while enabling horizontal motion. One roll unit is horizontally fixed while the other is
restrained via a compression spring, which allows it to resist a varying degree of horizontal load.

The pre-load on the movable roll can be adjusted up to a maximum of 20kN. The two motors that drive the
rolls are electronically synchronised through a variable speed controller, enabling the roll speed to be
continuously varied up to 14 rpm (approximately 0.14 m/s surface speed). The rolls have a centre-to-centre
distance ~,at zero gap setting) of I88mm and a width of 100mm. Both drive shafts are instrumented with
strain gauges to enable the roll torque to be measured. Additional sensors are provided to measure the
horizontal force on the stationary roll and the gap between the rolls. Clear glass is fitted to the sides of the
NCRC to facilitate viewing of the crushing zonc during operation and also allows the crushing sequence to
bc recorded using a high-speed digital camera.

Tests were performed on several types of rocks including granite, diorite, mineral ore, mill scats and
concrete. The granite and diorite were obtained from separate commercial quarries; the former had been
pre-crushed and sized, while the latter was as-blasted rock. The first of the ore samples was SAG mill feed
obtained from Normandy Mining's Golden Grove operations, while the mill scats were obtained from
Aurora Gold's Mt Muro mine site in central Kalimantan. The mill scats included metal particles of up to
18ram diameter from worn and broken grinding media. The concrete consisted of cylindrical samples
(25mm diameter by 25ram high) that were prepared in the laboratory in accordance with the relevant
Australian Standards. Unconfined uniaxial compression tests were performed on core samples (25mm
diameter by 25mm high) taken from a number of the ores. The results indicated strength ranging from 60
MPa for the prepared concrete up to 260 MPa for the Golden Grove ore samples.
All of the samples were initially passed through a 37.5mm sieve to remove any oversized particles. The
undersized ore was then sampled and sieved to determine the feed size distribution. For each trial
approximately 2500g of sample was crushed in the NCRC. This sample size was chosen on the basis of
statistical tests, which indicated that at least 2000g of sample needed to be crushed in order to estimate the
product P80 to within +0.1ram with 95% confidence. The product was collected and riffled into ten subsamples,
and a standard wet/dry sieving method was then used to determine the product size distribution.
For each trial, two of the sub-samples were initially sieved. Additional sub-samples were sieved if there
were any significant differences in the resulting product size distributions.
A number of comminution tests were conducted using the NCRC to determine the effects of various
parameters including roll gap, roll force, feed size, and the effect of single and multi-particle feed. The roll
speed was set at maximum and was not varied between trials as previous experiments had concluded that
there was little effect of roll speed on product size distribution. It should be noted that the roll gap settings
quoted refer to the minimum roll gap. Due to the non-cylindrical shape of the rolls, the actual roll gap will
vary up to 1.7 mm above the minimum setting (ie: a roll gap selling of l mm actually means 1-2.7mm roll
gap).
RESULTS
Feed material
The performance of all comminution equipment is dependent on the type of material being crushed. In this
respect, the NCRC is no different. Softer materials crushed in the NCRC yield a lower P80 than harder
materials. Figure 4 shows the product size distribution obtained when several different materials were
crushed under similar conditions in the NCRC. It is interesting to note that apart from the prepared concrete
samples, the P80 values obtained from the various materials were fairly consistent. These results reflect the
degree of control over product size distribution that can be obtained with the NCRC.
Multiple feed particles
Previous trials with the NCRC were conducted using only single feed particles where there was little or no
interaction between particles. Although very effective, the low throughput rates associated with this mode
of comminution makes it unsuitable for practical applications. Therefore it was necessary to determine the
effect that a continuous feed would have to the resulting product size distribution. In these tests, the NCRC
was continuously supplied with feed to maintain a bed of material level with the top of the rolls. Figure 5
shows the effect that continuous feed to the NCRC had on the product size distribution for the Normandy
Ore. These results seem to show a slight increase in P80 with continuous (multi-particle) feed, however the
shift is so small as to make it statistically insignificant. Similarly, the product size distributions would seem
to indicate a larger proportion of fines for the continuously fed trial, but the actual difference is negligible.
Similar trials were also conducted with the granite samples using two different roll gaps, as shown in
Figure 6. Once again there was little variation between the single and multi-particle tests. Not surprisingly,
the difference was even less significant at the larger roll gap, where the degree of comminution (and hence
interaction between particles) is smaller.

All of these tests would seem to indicate that continuous feeding has minimal effect on the performance of
the NCRC. However, it is important to realise that the feed particles used in these trials were spread over a
very small size range, as evident by the feed size distribution shown in Figure 6 (the feed particles in the
Normandy trials were even more uniform). The unilormity in feed particle size results in a large amount of
free space, which allow:s for swelling of the broken ore in the crushing chamber, thereby limiting the
amount of interaction between particles. True "choke" feeding of the NCRC with ore having a wide
distribution of particle sizes (especially in the smaller size range) is likely to generate much larger pressures
in the crushing zone. Since the NCRC is not designed to act as a "'high pressure grinding roll" a larger
number of oversize particles would pass between the rolls under these circumstances.

Roll gap
As with a traditional roll crusher, the roll gap setting on the NCRC has a direct influence on the product
size distribution and throughput of the crusher. Figure 7 shows the resulting product size distribution
obtained when the Aurora Gold ore (mill scats) was crushed at three different roll gaps. Plotting the PSO
values taken from this graph against the roll gap yields the linear relationship shown in Figure 8. As
explained previously, the actual roll gap on the NCRC will vary over one revolution. This variation
accounts for the difference between the specified gap setting and product Ps0 obtained from the crushing
trials. Figure 8 also shows the effect of roll gap on throughput of the crusher and gives an indication of the
crushing rates that can be obtained with the laboratory scale model NCRC.

Roll force
The NCRC is designed to operate with minimal interaction between particles, such that comminution is
primarily achieved by fracture of particles directly between the rolls. As a consequence, the roll force only
needs to bc large enough to overcome the combined compressive strengths of the particles between the roll
surlaces. If the roll force is not large enough then the ore particles will separate the rolls allowing oversized
particles to lall through. Increasing the roll force reduces the tendency of the rolls to separate and therefore
provides better control over product size. However, once a limiting roll force has been reached (which is
dependent on the size and type of material being crushed) any further increase in roll force adds nothing to
the performance of the roll crusher. This is demonstrated in Figure 9, which shows that for granite feed of
25-3 Imm size, a roll force of approximately 16 to 18 kN is required to control the product size. Using a
larger roll force has little effect on the product size, although there is a rapid increase in product P80 if the
roll force is reduced bek>w this level.

As mentioned previously, the feed size distribution has a significant effect on the pressure generated in the
crushing chamber. Ore that has a finer feed size distribution tends to "choke" the NCRC more, reducing the
effectiveness of the crusher. However, as long as the pressure generated in not excessive the NCRC
maintains a relatively constant operating gap irrespective of the feed size. The product size distribution
will, therefore, also bc independent of the feed size distribution. This is illustrated in Figure 10, which
shows the results of two crushing trials using identical equipment settings but with feed ore having
different size distributions. In this example, the NCRC reduced the courser ore from an Fs0 of 34mm to a
Ps0 of 3.0mm (reduction ratio of 11:1), while the finer ore was reduced from an Fs0 of 18mm to a Pso of
3.4mm (reduction ratio of 5:1). These results suggest that the advantages of using profiled rolls diminish as
the ratio of the feed size to roll size is reduced. In other words, to achieve higher reduction ratios the feed
particles must be large enough to take advantage of the improved nip angles generated in the NCRC.

Mill scats
Some grinding circuits employ a recycle or pebble crusher (such as a cone crusher) to process material
which builds up in a mill and which the mill finds hard to break (mill scats). The mill scats often contain
worn or broken grinding media, which can find its way into the recycle crusher. A tolerance to uncrushable
material is therefore a desirable characteristic for a pebble crusher to have. The NCRC seems ideally suited
to such an application, since one of the rolls has the ability to yield allowing the uncrushable material to
pass through.
The product size distributions shown in Figure 1 1 were obtained from the processing of mill scats in the
NCRC. Identical equipment settings and feed size distributions were used for both results, however one of
the trials was conducted using feed ore in which the grinding media had been removed. As expected, the
NCRC was able to process the feed ore containing grinding media without incident. However, since one
roll was often moving in order to allow the grinding media to pass, a number of oversized particles were
able to fall through the gap without being broken. Consequently, the product size distribution for this feed
ore shows a shift towards the larger particle sizes, and the Ps0 value increases from 4ram to 4.7mm. In spite
of this, the NCRC was still able to achieve a reduction ratio of almost 4:1.

Wear
Although no specific tesls were conducted to determine the wear rates on the rolls of the NCRC, a number
of the crushing trials were recorded using a high-speed video camera in order to try and understand the
comminution mechanism. By observing particles being broken between the rolls it is possible to identify
portions of the rolls which are likely to suffer from high wear and to make some subjective conclusions as
to the effect that this wear will have on the perlbrmance of the NCRC. Not surprisingly, the region that
shows up as being the prime candidate for high wcar is the transition between the flat and concave surfaces.
What is surprising is that this edge does not play a significant role in generating the improved nip angles.
The performance of the NCRC should not be adversely effccted by wear to this edge because it is actually
the transition between the fiat and convex surfaces (on the opposing roll) that results in the reduced nip
angles.
The vide() also shows that tor part of each cycle particles are comminuted between the flat surfaces of the
rolls, in much the same way as they would be in a jaw crusher. This can be clearly seen on the sequence of
images in Figure 12. The wear on the rolls during this part of the cycle is likely' to be minimal since there is
little or no relative motion between the particles and the surface of the rolls.

CONCLUSIONS
The results presented have demonstrated some of the factors effecting the comminution of particles in a
non-cylindrical roll crusher. The high reduction ratios obtained from early single particle tests can still be
achieved with continuous multi-particle feed. However, as with a traditional roll crusher, the NCRC is
susceptible to choke feeding and must be starvation fed in order to operate effectively. The type of feed
material has little effect on the performance of the NCRC and, although not tested, it is anticipated that the
moisture content of the feed ore will also not adversely affect the crusher's per[Brmance. Results from the
mill scat trials are particularly promising because they demonstrate that the NCRC is able to process ore
containing metal from worn grinding media. The above factors, in combination with the flaky nature of the
product generated, indicate that the NCRC would make an excellent recycle or pebble crusher. It would
also be interesting to determine whether there is any difference in the ball mill energy required to grind
product obtained from the NCRC compared that obtained from a cone crusher.