Chapter 52

ICE AS A COOLANT FOR DEEP MINES

T.J. Sheer	M.D. Butterworth
School of Mechanical Engineering University of the Witwatersrand, Johannesburg Private Bag 3, WITS, 2050 South Africa	CSIR Mining Technology P O Box 91230, Auckland Park, 2006 South Africa R. Ramsden Bluhm Burton Engineering (Pty) Ltd P O Box 786012, Sandton, 2146 South Africa

ABSTRACT

Ice is used in some deep Southern African mines as a heat transfer substance to transport refrigeration to underground working areas from ice-making installations located on the surface level. Three full-scale systems are currently in operation and economic studies indicate that further installations are possible in future. Two general types of ice have been used: `hard' ice in the form of dry fragments and `slush' ice resembling wet snow. The ice conveying systems are based on long pipelines connecting the ice plants to underground melting dams. Pipeline routes may include horizontal and vertical sections and may in future extend over total distances of up to five kilometres. The multi-phase flow through such pipelines has various unusual characteristics; extensive research has been required over the past twenty years to enable the design of reliable systems for conveying ice. The suppliers of ice-making systems have also carried out a great deal of development work over this period. This paper describes some of the research findings and engineering experience to date on ice cooling systems for mines.

KEYWORDS

Ice, cooling systems, deep mines, pneumatic conveying, refrigeration, pipelines, multiphase flow

INTRODUCTION

One of the most challenging engineering tasks in the establishment of very deep mines (3 000 m and more below surface) is the design of refrigeration and ventilation systems to provide suitable environmental conditions underground at acceptable cost. Heat typically needs to be removed from the underground areas of a large, deep mine at a rate of the order of 100 MW and discharged to the atmosphere. Chilled water systems are used to absorb and transport much of the heat, using refrigerators located either on surface or underground (and often both). When refrigerators are located on surface the warm water has to be pumped back from underground, at great cost. The advantage of using ice sent down from surface to absorb the heat load is that the quantity of water to be pumped back out of the mine can be reduced to about a quarter of that for an all-water cooling system, giving very large cost savings (Sheer, et al., 1984). This concept was initiated twenty years ago at East Rand Proprietary Mines Ltd (ERPM) (Sheer, et al., 1985) and a great deal of work has since been carried out to develop the technologies for both the manufacture and the transportation of ice for this unique application.

The ice fed into the conveying pipeline at the surface level of a mine needs to be in as dry a state as possible, consistent with the ice-making process employed, in order to minimize the return water pumping requirement. The design of a complete ice conveying system commences with a given ice flow rate, ice type, and pipeline route from the ice plant down to one or more underground melting dams. The route could include several vertical and horizontal sections. The following aspects require careful consideration by the designer: the pipe diameter, pipeline construction (including material, couplings, contraction arrangements, bends, and supports), a system to transport the ice on surface from the plant to the mineshaft and to inject it into the vertical pipeline in the shaft, and system control. The primary design criterion is the system reliability and availability. This necessitates the avoidance of flow blockages or pipeline failures.

ECONOMIC ASPECTS

A detailed evaluation of cooling generation systems for ultra-deep level gold mining has recently been carried out as part of a collaborative research programme (entitled `Deepmine') sponsored by certain South African mining companies. The full cost implications of using several alternative types of cooling systems were determined, for various mining depths down to 5 000 m below surface and for various mine designs. The study showed that the most economical form of cooling system for great depths is a hybrid system with a combination of underground and surface refrigeration machines. Because the maximum refrigeration capacity installed underground (presently in the form of conventional R-134a water chillers, although water vapour refrigeration could become attractive in future) would be limited by the combined heat rejection capacities of the air and water streams returning to surface, additional refrigeration would generally need to be installed on surface. The comparative evaluations confirmed that for depths below about 3 000 m ice systems would be the least expensive option for the surface plant at the present time, in spite of the high costs of making ice. Hydrolift systems (of which the three-chamber pipe feeder is one variant) also have attractive potential and are currently also being evaluated (Hegerman, 1997).

ICE MANUFACTURE

Current ice installations produce ice in one of two forms: `slush' ice (typically having 70 to 75 percent of ice by mass) formed by dewatering a dilute ice slurry (Ophir and Koren, 1999), or irregular `hard', essentially dry, particulate ice (Hemp, 1988). The feasibility of both types of ice has been demonstrated in large mine installations (see Table 1). The total costs (owning plus operating) of the two types of system generally seem to be comparable for a depth of 3 000 m. Slush ice plants have higher coefficients of performance (Ophir and Koren, 1999) but they incur higher water pumping costs because of the wetness fraction of the ice sent underground.

In the slush ice installation at the Mponeng shaft of Western Deep Levels Gold Mine, dilute ice slurries (about 17 percent ice crystals by mass) are first generated by means of vacuum evaporation at the triple point of water, dosed to have a salt concentration of about 0.35% (van der Westhuisen, 2000). The refrigerant is the water itself and the direct boiling process is inherently efficient, with no thermal resistance from any freezing surfaces in the evaporator. The dilute slurry is thereafter concentrated in gravity drain columns, the effectiveness of which is important (it is hoped to achieve ice mass fractions of over 80 percent in future). The development of large slush ice plants, such as those at Mponeng (where four 3 MW(R) machines are currently operating, each producing more than 800 t/d of concentrated slush ice), has required solutions to several challenging problems. The greatest of these has been the development of special water vapour compressors, with volumetric duties (for each Mponeng unit) of 320 m³/s over a compression ratio of 8:1 in two stages. Van der Westhuisen (2000) has described various aspects of the ice system performance at Mponeng.

Hard ice is made in more conventional installations with extensive freezing surfaces, using defrosting cycles for harvesting. The large installation at ERPM employs screw compressors with ammonia refrigerant (Hemp, 1988). Mains water is used to make the ice and one uncertainty in future applications would be whether using mine water as the feed would cause corrosion or fouling problems on the freezing surfaces. Water quality and brine management are also very important issues with vacuum ice machines, for other reasons (van der Westhuisen, 2000).

CONVEYING RESEARCH

A clear fundamental understanding of the multiphase flow in the various sections of a mine ice pipeline is required in order to formulate a pressure gradient model for design purposes and to design for flow reliability. As the ice, in whatever form, flows down the pipeline it melts progressively at a rate of typically 5 per cent of the initial mass of solid per 1 000 m of vertical descent. However air occupies the largest volume fraction in the pipeline and the flow may appropriately be analyzed as a two-phase flow of wet solids with air. Initial ice conveying research was based on the use of hard particulate ice while more recent work has considered slush ice. It is convenient to comment here on the research findings for these types of ice separately.

Hard Particulate Ice

Most of the existing research information was obtained from tests carried out on a pilot ice conveying installation located at ERPM (Sheer, 1995). This installation incorporated a pipeline that could be extended to depths down to 2 407 m below surface level, with a total length of up to 3 905 m (with two vertical and three horizontal sections). The ice, which had an initial average particle size of approximately 34 mm, was injected into the 136 mm inner diameter pipeline using a blow-through rotary valve. Initial tests showed that steel piping was unsuitable for ice conveying but an important finding was that plastic (in this case uPVC) piping was very successful. Wet ice is both highly cohesive and adhesive to carbon steel (and, to varying extents, to other metallic surfaces) and blockages resulted very readily. Hydrophobic surfaces such as uPVC, on the other hand, exhibit no adhesion with wet ice (and very low coefficients of sliding friction) and no blockages were experienced during a programme of over a hundred tests. Low-pressure uPVC piping could be used because the static pressures encountered during conveying were very low, given the absence of blockages. A typical maximum pressure was 220 kPa at the foot of the 1 770 m vertical section down the mine shaft; this was for an ice flow rate of 7.4 kg/s.

Photographic observations and pressure recordings at various points along the pipeline showed clearly that the nature of the flow changed significantly from one section of the pipeline to the next. If a particular configuration consisting of the first three sections of the pipeline is considered, the following sequence of flow patterns was observed.

Section 1. The first section was a 238 m horizontal pipeline from the ice plant to the top of the vertical mineshaft. Pneumatic conveying was employed to transport the ice along this section. Depending on the air velocity, the flow could be fully suspended dilute-phase (above 40 m/s), stratified dilute-phase (30-40 m/s), sliding clusters (27-30 m/s), or plug flow with nearly full-bore plugs (20-27 m/s). Similar successive flow regimes have been described frequently for a variety of conveyed materials (e.g. Molerus, 1981; Tsuji and Morikawa, 1982). In this case the `saltation' point occurred at approximately 30 m/s, this high value being attributable to the large particle sizes. The very low friction between wet ice and plastic permitted stable plug conveying down to approximately 20 m/s, after which blockages occurred shortly downstream of the rotary valve (where the air velocity was the lowest). Experimental results from these conveying tests have been published in the forms of phase-diagram plots and solids friction factors (Sheer, 1995). The main deficiency of the results is that the effect of pipe diameter was not adequately established, either in terms of the solids friction factor or the minimum conveying velocity. The important issue of the scaling of the pilot-plant results to full-size applications therefore remains partly unresolved.

Section 2. The second section was a 1 770 m vertical pipeline down the mineshaft, ending with a long-radius bend connecting to the following underground horizontal pipeline section. The downward flow was dilute-phase and the results from the large number of tests again allowed solids friction factors to be derived and expressed as a function of the Froude number, albeit only for a single pipe diameter. In some tests the air flow rate was controlled, being the same as that in the preceding horizontal section, while in others the air flows were allowed to be those naturally induced by the falling ice, by opening an air vent (Figure 1). The vertical solids friction factors were consistently less than those for the horizontal flow in the preceding section (Sheer, 1995). At the upper end of the vertical section the ice particles accelerate downwards until their velocity exceeds that of the air by the particle terminal velocity relative to the air, as indicated in Figure 1. Thereafter there is a quasi-equilibrium condition down the length of the pipeline, with the pressure and air density increasing and the air velocity and therefore the ice velocity decreasing. The pressure rise created by the downward-falling particles provides the driving force for the subsequent plug flow along the underground horizontal pipeline (Section 3).

Figure 1. Ice-air flow regimes (Sheer, 1995)

Figure 2 shows a typical set of pressure and velocity profiles along the complete pipeline, predicted on the basis of the experimental results and taking air compressibility into account. In this particular test the air vent valve at the surface level was closed. However very similar flow characteristics were measured in tests with the air vent valve open, when air was usually expelled from the vent. Plug flow always prevailed in the underground horizontal section in all tests. A very significant change to the flow occurs at the bend at the bottom of the vertical section, where the ice slides to a standstill. Plugs were formed immediately after this bend by collisions between the clusters of ice sliding out of the bend, air velocities being below the pick-up values for suspension flow at that point. Marjanovic, et al. (1987) observed similar flow patterns at the bottom of a vertical pipe in a laboratory system.

Figure 2. Typical pressure and velocity profiles along a three-section pipeline (Sheer, 1995)

Figure 3. Ice slug discharged underground

Section 3. The third section was a 630 m horizontal pipeline underground. The ice flowed along this section in the form of either full-bore or partial-bore plugs (the latter here called `slugs'), separated irregularly from each other by long air pockets. At low ice flow rates full-bore plugs formed but at rates higher than about 3 kg/s (depending upon the accompanying air flow rate) the plugs were longer and did not occupy the full bore (Figure 3). Large pressure fluctuations were recorded along the pipeline as full-bore plugs passed a sensor but the pressures were steadier in the slug-flow regime at the higher flow rates. The approach taken to model the flow of air-dragged slugs along this section was similar to that by Muschelknautz and Krambrock (1969). The resulting equation does not include a solids friction factor but does include the coefficient of sliding friction (measured to be approximately 0.02 for wet ice on uPVC). This approach is being reconsidered in current work, with the analysis rather being based on hydrodynamic lubrication principles. The shortcoming of the model developed for this type of flow is again that there is uncertainty about scaling the results to full-scale applications.

Slush Ice

Pilot testing was undertaken some years ago at the Western Deep Levels Gold Mine to investigate the flow of slush ice, containing about 30 percent of water by mass, through various pipelines (Ramsden, et al., 1994). A rotary valve was again used as the ice feeder and the pipeline incorporated a surface section 365 m long with pneumatic conveying (diameters of either 105 or 250 mm), a vertical section 1 200 m long (250 mm diameter) fed via an open funnel arrangement, and an
80 m horizontal section underground. The tests confirmed that reliable flow could be achieved using uPVC piping but detailed measurements were not recorded for the purpose of evaluating friction factors. The experience gained was nevertheless useful in the design of the full-scale installation for the same mine.

A comprehensive programme of laboratory flow tests is presently in progress, to obtain more detailed information on the mechanics of slush ice flowing with air through plastic pipelines of various diameters. Figure 4 shows an ice plug moving through a 52 mm diameter transparent pipe section downstream of a bend in this installation. The appearance is similar to that of plugs of particulate ice as they entered pipe section 3 in the pilot plant tests underground at ERPM.

Figure 4. Ice plug moving through 52 mm pipe

While experience to date suggests that the underground pipeline flow characteristics of slush ice are generally similar to those of hard particulate ice, a number of issues require resolution. These include quantitative descriptions of the flow conditions that would cause blockages to occur, and the formulation and verification of complete models for the prediction of pressure gradients. Experimental data are required for various pipe sizes in order to develop reliable scaling laws. One of the differences between flows of slush ice and particulate ice concerns the conditions under which transition occurs between different flow regimes. In particular, plug flow occurs more readily with slush ice than with hard ice. The transition point depends, however, upon the water content of the ice. (Notwithstanding the different transition characteristics, plug flow will always prevail underground with all ice in horizontal sections.) Figure 5, a static pressure record at one point in one of the laboratory pipelines, shows the much stronger pressure fluctuations that commence midway through a test when the inlet ice mass fraction is reduced from about 70 to 65 percent. These stronger fluctuations are indicative of plug, as distinct from dispersed-phase, flow.

MINE OPERATING EXPERIENCE

Significant quantities of ice have been conveyed from surface level to underground dams at four mine installations, in addition to the pilot plants already mentioned. Table 1 summarizes the salient data for all these installations up to the beginning of 2000. The flow rates given in this table are neither necessarily the total ice plant production capacities at the respective mines, nor the maximum carrying capacities of individual pipelines, but rather values for flows achieved routinely through single pipelines up to that time. (In the case of the ERPM pilot plant the maximum pipeline capacity could not be determined because of limitations in the ice feed rate).

With reference to Table 1, the installation at Harmony is no longer in operation because of closure of the shaft. The largest mine ice system is at ERPM, which has had ten years of successful operation. The total system capacity is 53 kg/s and the ice is normally conveyed underground to two different levels through two separate pipelines of equal diameter (the figures above represent the greatest recorded flow rate for a single pipeline). The transportation system between the ice machines and the mineshaft includes a combination of short screw conveyors, conveyor belts (200 m long) and, lastly, pneumatic conveying pipelines (50 m long) that connect directly to the vertical pipelines in the shaft. The uPVC pipes in the shaft have spigot and belled-end joints to accommodate thermal contraction and expansion. The joints are not airtight in this installation but there is no need to build up the pressure in the pipelines because the underground discharge points are shortly downstream of the final bends out of the shaft. In retrospect it is believed that flanged couplings would nevertheless have been more suitable, for various practical reasons.

The operation of this ERPM system has provided a wealth of valuable information, applicable to any other mine ice conveying system. One important issue in any such system is the mechanical design of the pipeline. As in the other systems mentioned in Table 1, it was found

Table 1. Particulars of ice conveying pipelines in mines (installations marked * are currently operating)

Mine	Type of ice	Flow rate (kg/s)	Pipe ID (mm)	Ice mass velocity (kg/s per m^2)	Total length (m)	Final depth (m)
ERPM (pilot)	Hard	7,4	136	500	3 900	2 400
Harmony	Hard	10	200	320	1 180	1 100
ERPM *	Hard	44	270	770	2 650	2 500
WDL (pilot)	Slush 70%	5	216	140	1 570	1 200
Mponeng *	Slush 70%	32	216	870	2 760	2 600
Selebi Phikwe*	Slush 65%	23	300	325	390	390

Figure 5. Pressure fluctuations in a pipeline caused by the onset of plug flow at 800 seconds

that ice plugs form along the pipeline (probably initiated in this case at intermediate deviations during the descent) and these are discharged violently at the end. The movement of the plugs causes the pipelines to shake severely and results in high impact forces at the pipe supports, especially if there is excessive clearance between the pipe and its supports. When using low-pressure plastic piping it is clearly important to minimize impact forces that can shatter the pipes, through careful support design. Another important issue concerns blockages. In this system (and at two other mines) blockages have occurred mainly at the ends of the pipelines because of overfilling of the underground ice dams into which they discharge; this can be avoided through suitable monitoring and control measures.

Mponeng mine operates the largest slush ice conveying system. Up to the beginning of 2000 the ice was fed into the top of a 216 mm inner diameter vertical pipeline through a funnel, into which four ice streams converged from four separate ice concentrators. The top of the funnel had a diameter of 1500 mm and the included angle was 20°. Occasional blockages occurred in the funnel, due to interference between the four ice streams (three were pneumatically conveyed, and the fourth was a gravity flow down a chute), until the feed pipes were rearranged. The piping used is again uPVC, with spigot and belled-end joints. The conveying system has operated reliably for some three years since inception. The only problems experienced were similar to those mentioned above for ERPM; there have been pipe breakages, probably due to unrestrained pipeline movement caused by the rather violent plug flow. Plugs formed during descent because of a diversion in the pipeline in the shaft. The ice-making installation will be expanded in the next phase to a capacity of 64 kg/s of ice (5 500 t/d). To accommodate this future flow the shaft pipeline was replaced during 2000 by a uPVC pipeline of 431 mm internal diameter. The pipe joints are again the bell-end and spigot type, with rubber O-rings for sealing.

The installation at Selebi Phikwe is interesting in that the slush ice is produced by vacuum evaporation, but using steam ejectors instead of mechanical compressors (Paul, et al., 1996). In this case steam was available at low cost from elsewhere. The conveying route is simple, comprising a single vertical pipe section.

CONCLUSIONS

Experience to date with ice cooling systems in Southern African gold mines demonstrates that ice can be manufactured in large quantities and that reliable flow can be sustained to great depths through long gravity-driven pipelines, with ice in the form of either hard particles or slush. The flow underground is mainly intermittent in nature with the ice plugs or slugs separated by long air pockets. Plastic piping such as uPVC must be used, with steel sheathing in areas where pipes are vulnerable to external physical damage. Because of the intermittent nature of the flow and the violent discharge at the end of a pipeline, particular care must be taken in the design of pipe supports to restrain the pipe and withstand dynamic loads. Flanged pipe couplings are preferable in order to maintain a smooth bore and to prevent air leakage, but provision must be made for thermal contraction and expansion. There must be no reductions in pipe cross-section, no diversions or off-takes, and no misalignment or protrusions at couplings. Large-radius bends must be used. The method presently favoured for injecting ice into the top of a vertical pipeline is to feed it into a funnel mounted directly onto the pipe. Further research is in progress to complete the formulation and verification of general mathematical models for multiphase flow through ice pipelines, and to incorporate these into general design guidelines.

REFERENCES

Hegerman, C., 1997, “Optimising the cooling of the Vaal Reefs No. 11 shaft underground environment and a description of refrigeration and energy recovery equipment used,” Journal of the Mine Ventilation Society of South Africa, Vol. 50, No. 1, Jan., pp. 18-24

Hemp, R., 1988, “A 29 MW ice system for mine cooling,” Proceedings, Fourth International Mine Ventilation Congress, A.D.S. Gillies, ed., AIMM, Brisbane, Australia, pp. 415-423

Marjanovic, P., Mills, D. and Mason, J.S., 1987, “A method for calculation of flow parameters for pneumatic conveying in pipelines,” Proceedings, Third International Conference on Pneumatic Conveying Technology, Pneumatech 3, Jersey, Channel Islands

Molerus, O., 1981, “Prediction of pressure drop with steady state pneumatic conveying of solids in horizontal pipes,” Chemical Engineering Science, Vol. 36, pp. 1977-1984

Muschelknautz, E. and Krambrock, W., 1969, “Simplified calculations for horizontal pneumatic systems conveying fine products at high loadings,” Chemie-Ing-Technik, Vol. 41, pp. 1164-1172

Ophir, A. and Koren, A., 1999, “Vacuum freezing vapor compression process (V.F.V.C.) for mine cooling, Proceedings, 20^th International Congress of Refrigeration, IIR/IIF, Sydney, Australia, September

Paul, J., Jahn, E., and Lausen, D., 1996, “Cooling of mines with vacuum ice,” Proceedings, FRIGAIR '96 Conference, SAIRAC, Kempton Park, March

Ramsden, R., Finlayson, R.M. and Butterworth, M.D., 1994, “Conveying slurry ice at Western Deep Levels,” Mechanical Technology, Dec., pp. 27-29

Sheer, T.J., Correia, R.M., Chaplain, E.J. and Hemp, R., 1984, “Research into the use of ice for cooling deep mines,” Proceedings, Third International Mine Ventilation Congress, M.J. Howes and M.J. Jones, ed., IMM, Harrogate, England, pp. 277-282

Sheer, T.J., Cilliers, P.F., Chaplain, E.J. and Correia, R.M., 1985, “Some recent developments in the use of ice for cooling mines,” Journal of the Mine Ventilation Society of South Africa, Vol. 38, pp. 56-59, 67-68

Sheer, T.J., 1995, “Pneumatic conveying of ice particles through mine-shaft pipelines,” Powder Technology, Vol. 85, pp. 203-219

Tsuji, Y. and Morikawa, Y., 1982, “Flow pattern and pressure fluctuation in air-solid two-phase flow in a pipe at low air velocities,” International Journal of Multiphase Flow, Vol. 8, pp. 329-341

Van der Westhuisen, L., 2000, “Vacuum ice technology, a deep level mining application,” Proceedings, FRIGAIR 2000 Congress, SAIRAC, Midrand, March

ACKNOWLEDGEMENT

The authors wish to acknowledge the support received from the `Deepmine' collaborative research programme for some of the work described here. Particular acknowledgement is made to Mr R McGarry and Miss C Y Chan, who are closely involved with the current research into ice conveying and who provided Figures 4 and 5.

2

I SZKOŁA AEROLOGII GÓRNICZEJ 1999

5

360

PROCEEDINGS OF THE 7^TH INTERNATIONAL MINE VENTILATION CONGRESS

361

ICE AS A COOLANT FOR DEEP MINES

---