Optimization of High-Pressure Grinding Roll (HPGR) Performance in an Industrial-Scale HPGR/Tower Mill Comminution Circuit

Abstract

The integration of high-pressure grinding roller (HPGR) with pre-concentration techniques and stirred mills is recognized for its energy efficiency. Studies have suggested that the feed with a P₈₀ around 1 mm is acceptable for stirred mills or coarse particle flotation. Nonetheless, published experimental data characterizing the comminution behavior of single-stage HPGR circuits configured with a 1 mm screen aperture remain scarce. Moreover, extant research remains confined to laboratory scale. Consequently, critical performance metrics, including production capacity, screening efficiency, and process continuity, have not been substantively documented in the literature. In this paper, the HPGR performance in an industrial-scale HPGR/tower mill comminution circuit was assessed and optimized by laboratory and industrial tests. The research meticulously analyzed the impact of feed rate on the industrial-scale flip-flow screen and HPGR performance and found that the HPGR featuring two studded rolls with a diameter of 800 mm and a width of 400 mm, operating in a reverse classification circuit with a scalped feed by a 14.64 m² flip-flow screen while running continuously 24 h per day, is capable of producing a −1 mm comminution product suitable for tower mill feed. Under the optimal operating conditions identified, it achieved a specific energy consumption of 4.57 kWh/t with a feed rate of 27.08 t/h.

1. Introduction

Comminution processes, especially grinding, have been identified as highly energy-intensive and energy-inefficient, accounting for up to 1.8% of global electricity generation [1] and 60% of total energy consumption in processing plants, yet achieving energy efficiencies as low as 1% [2]. There is every indication that the energy cost of comminution will continue to rise in the coming years. The reasons for this lie in the continuously increasing cost of generating electricity and in the perceptible trend towards finer grinding of ores on an increasingly larger scale for mineral liberation from lower-grade ores and for producing ultrafine particulates for emerging materials technologies. Consequently, optimizing and controlling energy usage in particle size reduction units is crucial [3].

HPGRs have been widely recognized as an energy-efficient alternative to conventional tertiary or quaternary crushers [4,5,6]. HPGRs offer several potential benefits, including reduced energy consumption, increased throughput, enhanced mineral liberation, particle weakening, and the induction of microcracks at the grain boundaries [7,8,9]. These material modifications contribute to lower energy requirements, higher grades, and improved recovery rates in the downstream process [10,11,12,13,14]. Furthermore, energy efficiency can be significantly enhanced by replacing traditional primary grinding equipment with HPGR systems [15].

A tower mill is a kind of wet vertical stirred grinding equipment utilizing steel balls or ceramic grinding media. It has been utilized in fine and ultra-fine grinding for many years [16]. To enable the use of the energy-efficient vertical stirred mill, the transfer size must be controlled by closing the upstream with a screen, which allows other energy-efficient upstream technologies to be utilized, such as HPGRs. Pilot test work was carried out by [17] on a batch ball mill and a pilot-scale vertical mill to assess the energy efficiency factor. The results revealed that the vertical mill achieved a 35% higher efficiency compared to the batch ball mill when processing iron and copper samples with feed sizes of up to 1.57 mm and 1.13 mm, respectively. These findings demonstrated that the stirred mill performs effectively for coarser feed sizes. Regarding pre-concentration, significant advancements in coarse particle separation technology have been documented. Studies suggest that the upper limit of flotation may be extended up to 1 mm [18,19,20,21].

The improved utilization of high-pressure grinding rolls (HPGRs) and stirred mills for coarser grinding is being actively explored, particularly when integrated with pre-concentration [1]. Ref. [22] proposed the very first concept in which two stages of the HPGR produce material with a P₈₀ of 1.5 mm, which can then be fed directly to a Vertimill. Simulation results presented in the study demonstrated energy savings exceeding 45% when employing high-intensity blasting, HPGRs, and stirred milling circuits, compared to conventional blasting/SAG/ball mill circuits. Similar findings were reported by [10,23], who concluded that the two-stage HPGR–stirred mill circuit is more energy-efficient than HPGR–ball mill circuits, cone crusher–ball mill circuits, and SABC–ball mill circuits. However, studies of a one-stage HPGR closed with around 1 mm screen apertures are very rare. In addition, these studies did not investigate the impact of throughput, as they were conducted under laboratory-scale conditions. While laboratory-scale tests are valuable for pre-feasibility assessments, they may not fully represent industrial-scale operations [3]. Pilot-scale HPGR testing is considered the only available option to estimate the specific energy, particle size reduction, and circulating load for a closed-circuit HPGR operation [2]. Ref. [24] provided a brief overview of the performance of an industrial-scale HPGR–tower mill comminution circuit. In this study, the performance of an industrial-scale HPGR–stirred mill circuit was investigated in detail.

To provide technical support for the promotion and application of the HPGR/tower mill comminution circuit, the HPGR performance in an industrial-scale HPGR/tower mill comminution circuit was assessed and optimized in this paper. First, laboratory-scale HPGR pressure force-sensitive tests were conducted in an open circuit. On this basis, reverse and forward classification circuits with a 1 mm screen aperture were investigated. Finally, and most importantly, the effect of feed rate on the screening efficiency of an industrial-scale flip-flow screen, recycled load, specific pressure, roll gap, specific energy consumption, and fines generated (%−75 μm) were studied in detail in an industrial-scale HPGR. The performance of the tower mill and pre-concentration in the circuit will be reported in a future paper.

2. Materials and Methods

2.1. Sample and Characterization

Bulk iron–zinc ore samples were sourced from Sujia Mine, located in Heilongjiang province in China. The iron–zinc ore deposit comprises four distinct ore bodies, and the ore dressing plant typically processes a blend of these ore bodies. During the tests conducted in this study, the ore dressing plant was fed with material from the III^# Zinc ore body. The chemical composition and XRD pattern of the representative feed sample are provided in

Table 1. Chemical composition of raw ore (mass fraction, %).

Elements	Fe2O3	S	SiO2	Zn	Cu	Pb
Content	21.31	12.55	28.35	3.52	0.40	0.83

and Figure 1, respectively.

The chemical analysis results in Table 1 show that the iron–zinc ore consisted of 21.31 wt.% Fe₂O₃, 12.55 wt.% S, 28.35 wt.% SiO₂, and 3.52 wt.% Zn. Combining the XRD spectra in Figure 1 and the chemical compositions provided in Table 1, it can be concluded that the major minerals present are pyrite, sphalerite, quartz, magnetite, pyrrhotite, and chlorite.

2.2. Bond Work Index

The homogenized sample was crushed, and a series of −3.2 mm fractions were prepared according to the standard Bond Work Index test procedure. Bond ball mill locked-cycle tests were conducted using a standard laboratory-scale Bond mill (305 mm in diameter) with a standard Bond ball charge at a closing screen size of 0.074 mm. The work index (WI), expressed in kWh/t, was calculated using the equation proposed by Bond [25]. As shown in

Table 2. BMWI test results for raw ore.

Bulk Density	F80	P80	Closing Screen Size	BBWI
2.07 t/m3	1820 μm	51.18 μm	74 μm	16.06 kwh/t

, the Bond Ball Work Index (BBWI) of the sample was determined to be 16.06 kWh/t.

2.3. Laboratory-Scale Tests

A laboratory-scale WGM-3516 HPGR (manufactured by Shenyang Shengshi Wuhuan Science and Technology Co., Ltd., Shenyang, China) (Figure 2) with a 350 mm roll diameter and 160 mm roll width and a roll speed of 0.4 m/s was used in the laboratory-scale tests. The specifications of the unit are listed in

Table 3. Laboratory-scale WGM-3516 HPGR.

Item	Description
HPGR model	WGM-3516
Roll diameter	350 mm
Roll width	160 mm
Initial gap (gap zero)	6 mm
Maximum gap	15 mm
Wear surface	Studded
Motor power	2 × 18.5 kW
Maximum specific pressure	6 N/mm2
Variable speed drive	0~0.4 m/s

.

In each test, 25 kg of the ore sample was accurately weighed and utilized. Open-circuit tests were conducted at four specific pressures of 3.0, 3.5, 4.0, and 4.5 MPa, respectively. In closed-circuit tests, the ore sample was passed through the HPGR, followed by dry screening with a 1 mm screen size. The fresh feed and the oversize material were fed into the next cycle, where fresh materials were equal to the weight of the previous undersize fractions. This grinding and screening procedure was repeated gradually until the undersize weight was equal to the fresh feed in the same cycle. After about six locked cycles were performed, the −1 mm products of the last three cycles were combined, homogenized, and sampled for size analysis.

2.4. Industrial-Scale HPGR Tests

Industrial-scale HPGR tests were conducted at the Harbin Xiaoling iron–zinc mineral processing plant, utilizing an HPGR unit with a roll diameter of 800 mm and a roll width of 400 mm. The comminution circuit comprised several key components: a jaw crusher operating in an open circuit, a reverse-closed cone crusher paired with a circular vibrating screen, a closed HPGR circuit integrated with a vibrating flip-flow screen and a closed tower mill circuit equipped with hydrocyclones. The circular vibrating screen was used to achieve an aperture size of −12 mm for the HPGR feed, while the configured vibrating flip-flow screen produced an undersize product with 90% of particles measuring less than 1 mm. The flowsheet and main equipment are presented in

Table 4. Main equipment.

Equipment Type	Quantity	Specification	Power (kW)
Primary jaw crusher	1	C80	75
Secondary cone crusher	1	GP100M	90
Circular vibrating screen	1	YA1836	11
HPGR	1	WGM8040	2 × 110
Vibrating flip-flow screen	1	SSMS2461	33
Tower mill	1	WTM-355	355
Cyclones	1	FXΦ350-4	-
Cyclone feed pump	1	150JZ	55

and Figure 3, and the industrial-scale HPGR, vibrating flip-flow screen, and tower mill are shown in Figure 4.

The jaw crusher, cone crusher and circular vibrating screen were supplied by Shunda Mining Group, located in Shenyang, China. The HPGR and Tower mill were manufactured by Shenyang Shengshi Wuhuan Science and Technology Co., Ltd., located in Shenyang, China. The vibrating flip-flow screen was provided by Selm Technology Co., Ltd, located in Shenyang, China. The cyclones and Cyclone feed pump were produced by HaiWang Hydro cyclone Co., Ltd., located in Weihai, China.

The comminution circuit incorporated a WGM-8040 HPGR unit, as illustrated in Figure 4a, featuring two studded rolls with a diameter of 800 mm and a width of 400 mm. Detailed specifications of the unit are presented in

Table 5. HPGR specifications.

Item	Description
HPGR model	WGM-8040
Roll diameter	800 mm
Roll width	400 mm
Initial gap (gap zero)	8 mm
Maximum gap	20 mm
Wear surface	Studded
Motor power	2 × 110 kW
Maximum specific pressure	8.5 N/mm2
Variable speed drive	Up to 40 rpm (1.55 m/s)

. When operated, the initial specific pressure and initial roll gap were set at 4.0 MPa and 8 mm, respectively. The HPGR’s key process operating parameters, such as hopper level, roll speed, working gap, hydraulic pressure, and fixed and floated power, were monitored and recorded by automatic systems.

The vibrating flip-flow screen, extensively employed for screening moist fine-grained minerals [26], as shown in Figure 4b, is composed of a screen frame, a floating screen frame, rubber shear springs, supporting springs, supporting frames, and elastic sieve mats [27]. The specifications of the unit are provided in

Table 6. Vibrating flip-flow screen specifications.

Item	Description
Screen model	SSMS2461 (single layer)
Screen area	14.64 m2
Screen size	1 mm × 10 mm
Power	30 kW
Speed	800 rpm

.

3. Results and Discussion

3.1. Laboratory-Scale Tests

3.1.1. Open HPGR Circuit

Specific pressure sensitivity tests were conducted under open-circuit conditions, as illustrated in Figure 5. Samples were comminuted by the HPGR at specific pressures of 3.0, 3.5, 4.0, and 4.5 MPa, and the feeding and product particle size distribution (PSD) at different specific pressures are presented in Figure 6 and Figure 7, respectively. For quantitative analysis of the PSD characteristics, the Rosin–Rammler distribution function [28], Equation (1), was employed to mathematically describe the PSD.

$$y=1-exp\left[-{\left({\displaystyle \frac{x}{x^{\prime }}}\right)}^n\right]$$

(1)

where y represents the cumulative percentage undersize mass fraction distribution function, and x denotes particle size (x ≥ 0). The parameter x′ corresponds to the characteristic particle size at which 63.1% of the particles pass through the sieve, serving as an indicator of the average particle size. The exponent n, known as the non-uniformity coefficient, quantifies the degree of particle size distribution (PSD) heterogeneity [28,29].

It is observed in Figure 6 and Figure 7 that size reduction occurred as the coarse fraction decreased and the fine portion increased. As specific pressure increased up to 4.0 MPa, the −12 + 8 mm fraction changed from 24.17% to 2.14%, with a net decrease of 22.03%; the −0.074 mm fraction changed from 6.35 to 14.84%, with a net increase of 8.49%; and the P₈₀ changed from 9.14 mm to 3.40 mm, with a net decrease of 5.74 mm. As specific pressure continued to increase from 4.0 MPa to 4.5 MPa, the −12 + 8 mm fraction changed from 2.14% to 1.28%, with a net decrease of only 0.86%; the −0.074 mm fraction changed from 14.84% to 15.93%, with a net increase of 1.09%; and the P₈₀ changed from 3.73 mm to 3.42 mm, with a net decrease of 0.31 mm.

As illustrated in Figure 7, the Rosin–Rammler function exhibited an excellent fit with the experimental data, demonstrating its capability to accurately describe and predict the particle size distribution (PSD) across various specific pressure levels. The analysis revealed that as the specific pressure increased from 3.0 MPa to 4.0 MPa, the characteristic particle size (x′) decreased significantly from 3.56 mm to 2.60 mm. However, a further increase in specific pressure to 4.5 MPa resulted in only a marginal reduction in average particle size, from 2.60 mm to 2.44 mm. Concurrently, the non-uniformity coefficient (n) showed a slight decrease from 0.9107 to 0.8246 as the specific pressure increased from 3.0 MPa to 4.5 MPa, indicating a marginally wider PSD. This wider PSD can be attributed to the generation of finer particles, despite the reduction in coarse particles. The parameters of the Rosin–Rammler function consistently demonstrated that higher specific pressures lead to smaller average particle sizes and more uniform particle size distributions.

In mineral comminution processes, one of the most critical parameters is the mass-specific comminution energy (E_SE) associated with t₁₀, which is defined as the cumulative mass percentage of the product passing 1/10th of the initial feeding size [30]. In this study, the net specific energy (E_NSE) for the HPGR operating in the open comminution circuit was calculated as the power input integrated over the grinding time (t) and divided by the mass of feeding (m) [31], as shown in Equation (2).

$$E_{NSE}={\displaystyle \frac{P_G-P_N}{m}}$$

(2)

where E_NSE is the net specific energy (kWh/t), P_G is the gross power draw (kW), P_N is the no-load power draw (kW), and m is the dry mass of HPGR feed (t/h).

Figure 8 displays the trends in t₁₀ and E_sc at four specific pressure levels. At specific pressures of 3.0 MPa, 3.5 MPa, 4.0 MPa, and 4.5 MPa, the corresponding t₁₀ values were 36.22%, 42.55%, 47.4%, and 49.6%, respectively, while the E_sc values were 1.74, 1.87, 2.02, and 2.23, respectively. It is evident that when the specific pressure exceeded 4.0 MPa, the rate of the increase in t₁₀ slowed significantly, whereas the E_NSE showed a notable rise. Regarding the relationship between specific pressure and Ecs, further increasing the specific pressure led to an increase in the HPGR shaft torque, which subsequently increased the total power drawn from the motor [32].

A quantitative relationship exists between comminution energy and the particle size of breakage products in the comminution process [33]. In this study, the modified Weibull distribution model (Equation (3)) [34] was employed to describe the breakage index t₁₀ (%) in relation to the mass-specific impact energy, presented in Figure 9.

$$t_{10}=M\left\{1-exp\left[-f_{mat}·X·\left(E_{NES}-E_{min}\right)\right]\right\}$$

(3)

where M (%) represents the maximum t₁₀ for a material subject to breakage, X (m) is the initial particle size, E_NES (J kg⁻¹) is the mass-specific impact energy, and E_min (J kg⁻¹) is the threshold energy.

As depicted in Figure 9, the high R² value indicates that the model provides an excellent fit to the experimental data. When the net specific energy exceeded 2.02 kWh/t (corresponding to a specific pressure of 4.0 MPa), the growth rate of t₁₀ (i.e., the slope of the curve in the plot) decreased significantly.

In related studies, refs. [35,36,37] observed the same phenomenon during interparticle breakage tests and attributed the fracture kinetics to a distinct feature of the interparticle breakage mechanism. As the specific pressure increases, due to the gradual filling of void spaces in the bed of particles by fine-produced particles and as a result of the absence of any void space, the hydrostatic state is reached in the bed, and breakage comes to a “halt” [36].

To sum up, increasing the specific pressure from 3.0 MPa to 4.0 MPa resulted in efficient size reduction. However, when the specific pressure exceeded 4.0 MPa, the rate of size reduction diminished significantly, while the specific energy consumption increased sharply. Therefore, a specific pressure of 4.0 MPa was selected for the subsequent closed-circuit operations.

3.1.2. Closed HPGR Circuit

Instead of an open HPGR circuit, a closed HPGR circuit incorporating classification (screening) is employed to provide a defined product with a fixed top size and a higher fines content [5]. From an energy efficiency perspective, if the feed contains minimal material of final product size, it is generally more advantageous to feed it directly to the HPGR, as the classifying device would otherwise return all material to the HPGR, placing unnecessary load on the screen. Conversely, if the feed contains a significant amount of fines and the classification process is efficient, pre-classifying the material can be beneficial. In this study, given that approximately 30% of the fresh feed consisted of a minus 1 mm fraction, as shown in Figure 7, a reverse comminution circuit was recommended.

Figure 10 illustrates the mass balance of the two closed HPGR circuit configurations: (a) forward classification, where the fresh feed is directed straight to the HPGR, and (b) double classification, combining both scalped and reverse classification, where the fresh feed is first sent to the screen [38]. PSDs of the undersize for closed HPGR circuit tests are presented in Figure 11.

As shown in Figure 10, the double classification configuration (Figure 10b) demonstrated significant advantages over the forward classification case (Figure 10a). Specifically, the recycling load decreased by 30.84%, from 96.45% to 65.61%, and the throughput was reduced by 58.41%, from 196.45% to 138.08%, due to 27.53% of fine material bypassing the HPGR. Additionally, Figure 11 reveals that the P₈₀ of the undersize material was 0.64 mm for the forward classification circuit and 0.66 mm for the double classification circuit, indicating no significant difference in product fineness.

The results presented in Figure 10 and Figure 11 suggest that scalping the feed can reduce the required HPGR size, lower capital costs, and decrease comminution energy consumption by removing a portion of fines from the HPGR feed. Therefore, reverse classification was selected for the subsequent industrial tests.

3.2. Industrial Test Results

Attempts at improving comminution machines have generally been directed towards increasing performance efficiency, particularly increasing the throughput rate and reducing energy consumption [39]. Throughput stands as a pivotal metric for assessing comminution circuits, directly influencing both energy consumption and the economic efficiency of beneficiation plants. As mentioned earlier, few studies exist on the HPGR/tower mill comminution circuit, and there is still a lack of assessment of capacity.

Furthermore, within the HPGR/tower mill comminution circuit, the HPGR-crushed product sizing (−1 mm) presents a significant caking tendency. To mitigate this issue, no surge bin will be placed between the HPGR and the tower mill, and the upstream HPGR throughput will be adjusted according to the downstream tower mill’s capacity. Consequently, investigating the impact of feed rate on HPGR and flip-flow screen performance becomes imperative.

In the industrial test, the fresh-feed rate of the HPGR was regulated using a disc feeder. At disc feeder frequencies of 35, 40, 45, 50, 55, and 60 Hz, the corresponding fresh-feed rates were 18.98, 21.67, 24.38, 27.08, 29.79, and 32.50 tones per hour (tph), respectively. The industrial-scale circuit configurations and sampling positions are illustrated in Figure 12.

In closed HPGR circuit configurations, the total feed to the HPGR consists of fresh feed and recycled coarse material from the overflow of the flip-flow screen. However, in practice, due to the particle size characteristics and the limited screening time, fine particles cannot be fully separated into the underflow. Therefore, screening efficiency is used to characterize the actual screening performance [40].

The screening efficiency (E), recycled load (C), and HPGR throughput (Q₄) can be expressed in terms of the measured undersize mass fractions in each stream and calculated using Equations (4)–(6) [41].

$$E={\displaystyle \frac{{\beta }_2-{\beta }_4}{{\beta }_2(1-{\beta }_4)}}$$

(4)

$$C={\displaystyle \frac{1-{\beta }_{1}E}{{\beta }_{5}E}}$$

(5)

$$Q_4=C{Q}_1$$

(6)

where β₂ and β₄ represent the proportions of screen underflow for the 1 mm aperture screen in the feed and overflow for the flip-flow screen, respectively. β₁ and β₅ represent the proportions of screen underflow for a 1 mm aperture screen in the fresh feed and the comminution product of the HPGR, respectively, as shown in Figure 12. Q₁ represents the HPGR fresh-feed rate.

Screening efficiency and recycled load, calculated using Equations (4) and (5), are presented in Figure 13 for the feed frequency sensitivity tests.

The results indicate that as the feed frequency increased from 18.98 tph (35 Hz) to 32.50 tph (60 Hz), the screening efficiency of the flip-flow screen decreased from 86.45% to 60.52%, while the recycled load increased from 230.23% to 336.98%. This trend is attributed to the increased processing capacity per unit area of the screen at higher feed rates, which led to a thicker material layer. This made it more difficult for fine particles to pass through the layer and reach the screen surface [40]. Consequently, fine materials overflowed with coarse products and were returned to the HPGR as recycled material, further increasing the recycled load.

Figure 14 illustrates the relationship between the fresh-feed rate and the back-calculated throughput in the feed frequency sensitivity tests.

As shown in Figure 14, when the fresh-feed rate increased by 66.66% from 18.98 tph (20 Hz) to 27.08 tph (50 Hz), the throughput increased by 58.91% from 69.35 tph to 107.51 tph. However, the fresh feed increased by 20.01% from 27.08 tph (50 Hz) to 32.50 tph (60 Hz), with the throughput increasing by 55.03% from 69.35 tph to 107.51 tph due to the decrease in screening efficiency and the increase in recycled load.

Combining the insights from Figure 13 and Figure 14, it can be inferred that the small specification of the flip-flow screen and its mismatch with the HPGR led to a rapid decline in screening efficiency as the fresh-feed rate increased. This resulted in a higher recycled load and worsened comminution performance. To mitigate these negative effects, a larger HPGR model is required to accommodate the reduced screening efficiency and improve overall circuit performance.

To investigate the influence of feed rate on specific pressure and the operational gap, a constant hydraulic pressure of 15 MPa (corresponding to an initial specific pressure of 4.0 MPa) was maintained throughout all industrial trials. The variations in specific pressure and operational gap for each test are illustrated in Figure 15.

As evident from Figure 15, an increase in the fresh-feed rate resulted in a corresponding rise in both specific pressure and operational gap. When the feed frequency was increased from 35 Hz to 60 Hz, the specific pressure increased from 3.5 MPa to 4.1 MPa, while the operational gap expanded from 14.03 mm to 28.53 mm. Notably, the operational gap exhibited a trend similar to that of the specific pressure, as depicted in Figure 15. However, these findings deviate from the results reported by [42,43], where HPGRs operating under a constant feed rate demonstrated a reduction in operational gap with increasing pressing force. This discrepancy highlights the significant role of feed rate variability in influencing the operational dynamics of HPGRs.

The operational gap is inherently not directly controllable, as it results from the dynamic interaction between the particle bed and the hydraulic system. The roller shifting velocities and positions can be determined using the following equations [44]:

$$\ddot{x}(t_n)={\displaystyle \frac{f\left(t_n\right)+c\dot{x}\left(t_n\right)+kx\left(t_n\right)}{m}}$$

(7)

$$\dot{x}\left(t_{n+1}\right)=\dot{x}\left(t_n\right)+\ddot{x}\left(t_n\right)∆t$$

(8)

$$x\left(t_{n+1}\right)=x\left(t_n\right)+\ddot{x}\left(t_{n+1}\right)∆t$$

(9)

Here, $f\left(t_n\right)$ represents the total force acting on the floating roller at the current time step, while $x\left(t_{n+1}\right)$, $\dot{x}\left(t_{n+1}\right)$, and $\ddot{x}\left(t_n\right)$ denote the roller position, velocity, and acceleration at the next time step, respectively. Additionally, $∆t$ is the time step, and m is the mass of the floating roll.

Ref. [45] compared the comminution behavior of HPGRs and jaw crushers, highlighting that HPGRs exhibit time-dependent dynamic behavior influenced by loading conditions. The compressive force exerted on the roller is fed back into the model for the next update, causing the roller to retract if internal forces exceed a certain threshold. Consequently, even under a constant hydraulic pressure setting, an increase in fresh-feed rate leads to higher specific pressure and a larger operational gap. A similar argument was also proposed by [46], further supporting this phenomenon.

Energy, as a critical variable in comminution, plays a central role in various comminution models, serving either as an input or output parameter [47]. Compared to conventional comminution equipment, HPGRs offer significant technological and economic advantages, including more efficient particle breakage and lower energy consumption per unit mass of processed feed material across the entire crushing and grinding circuit [48]. These energy savings in downstream comminution are primarily attributed to particle weakening and the increased generation of fine particles, particularly those below 75 μm [49], which surpasses the performance of traditional comminution machinery [50]. The effect of fresh-feed rate on HPGR circuit-specific energy [51], calculated using Equation (10), and the generation of fines (%−75 μm) were investigated, as illustrated in Figure 16.

$$E_{sp}={\displaystyle \frac{P}{Q}}$$

(10)

where E_sp is the specific energy consumption (kWh/t), P is the consumed power (kW), and Q is the fresh-feed rate (t/h).

The results in Figure 16 demonstrate that increasing the fresh-feed rate initially reduced E_sp from 5.86 kWh/t to 4.57 kWh/t at a feed frequency of 55 Hz, after which E_sp slightly increased to 4.70 kWh/t at 60 Hz. Conversely, the fines generated (%−75 μm) exhibited an inverse trend, reaching a maximum of 35.95% at 50 Hz before declining slightly at higher feed frequencies.

The results from Figure 16 also suggest that increasing the fresh-feed rate to a proper level had a positive effect on E_sp reduction in HPGR and subsequent grinding stages, which could be attributed to interparticle breakage [52]. During particle bed compression, smaller particles fill the voids between larger particles, facilitating direct energy transfer and size reduction within the particle bed rather than through surface contact crushing between the rollers.

However, as shown in Figure 16, exceeding a certain fresh-feed-rate threshold negatively impacts both E_sp and fines production. This may be due to an increase in fresh-feed rate directly causing an increase in specific pressure, then increasing the total power drawn from the motor, and consequently having an adverse effect on E_sp. Additionally, excessive specific pressure can lead to over-compression, resulting in the formation of larger cakes that hinder fines production and screening efficiency. Similar observations have been reported by [10,48].

With a throughput of 27.08 t/h (50 Hz), the circuit has been operating for more than 10,000 h, confirming that an HPGR operating with a closed-circuit flip-flow screen aperture of approximately 1 mm can efficiently provide the feed for a downstream stirred mill in an HPGR/tower mill comminution circuit.

4. Conclusions and Outlook

This study presents an industrial-scale application where an HPGR produced a −1 mm comminution product for a tower mill. To comprehensively evaluate HPGR performance, the article examined multiple factors, including screening efficiency, recycled load, specific pressure, roll gap, specific energy consumption, and fines generation (%−75 μm). The key findings are summarized as follows:

(1)

Higher specific pressures resulted in finer products and increased energy consumption. A specific pressure of 4.0 MPa was identified as the optimal balance, achieving reasonable product fineness and moderate energy consumption.

(2)

Compared to forward classification, a reverse classification circuit with a scalped feed to the laboratory-scale HPGR reduced the recycling load by 30.84%, and the throughput was reduced by 58.41%.

(3)

An appropriate increase in feed rate reduced screening efficiency but enhanced HPGR performance, particularly in terms of energy efficiency. Increasing the fresh-feed rate from 18.98 tph to 32.50 tph initially reduced specific energy consumption from 5.86 kWh/t to 4.57 kWh/t at 29.79 tph, after which it slightly increased to 4.70 kWh/t.

(4)

With a throughput of 27.08 t/h, the circuit has operated for more than 10,000 h, confirming that an HPGR operating with a closed-circuit flip-flow screen aperture of approximately 1 mm can efficiently provide the feed to a downstream stirred mill in an HPGR/tower mill comminution circuit.

To further promote their application, future research will focus on the performance of the tower mill and pre-concentration processes within the industrial-scale HPGR/tower mill comminution circuit.