Chapter 108

THE ROLE OF WATER CHILLING MACHINES IN OPTIMISING PERFORMANCE OF EXISTING MINE COOLING SYSTEMS

M. Bailey-McEwan

School of Mechanical Engineering University of the Witwatersrand, Johannesburg Private Bag 3, 2050 Wits South Africa

ABSTRACT

The duties of existing mine water chilling installations tend to change as mining progresses. Sometimes, the water chilling machines cannot yield their full capacity at the new duties, being limited by their existing control philosophy or arrangement in the installation. Guidelines for altering control philosophy and machine arrangement to maximise water chilling capacity are suggested and illustrated in two simulation studies. In the first, where four machines were operating on part-duty, an altered “maximum-load” control philosophy was proposed. Apart from predicting increases of 46 per cent in available chilling capacity and 14 per cent in overall COP, one machine was freed for maintenance during normal working hours. In the second study, where four machines were arranged in series-parallel in their chilled water circuit, an alternative arrangement was proposed to permit delivering more chilled service water if mining operations required it. This “junction series-parallel arrangement” of the same machines was predicted to deliver 34 per cent more chilled service water over 24 hours in mid-summer. Control philosophy and machine arrangement should always permit maximum flexibility and effectiveness of existing capital equipment, which may have been designed and purchased for considerably different duties.

KEYWORDS

Mine cooling, water chilling installations / machines, installation layout, machine control, cooling capacity

introduction

A decade ago, it was estimated that the costs of ventilating and cooling some deep South African gold mines could approach 30 per cent of total working costs (Chamber of Mines of South Africa, 1991). So there are strong incentives to optimise the performance of ventilating and cooling systems in deep, hot South African gold and platinum mines. This paper examines the role of water chilling machines - the most complicated and costly components of mine cooling systems - in such optimisation. These machines are generally located in centralised water chilling installations either underground or on surface.

Attention is focused on existing water chilling machines in existing installations. Apart from diurnal and seasonal variations, the duties of such machines may increase or decrease substantially, suddenly and lastingly due to the unpredictable nature of mining (van der Walt, 1979). The resulting altered duties are liable to differ considerably from the original design ones for which the machine manufacturer specified optimal performance at the time of ordering. Obviously, maximum flexibility in existing installations to adapt to such changes in duties is thus desirable.

Two methods of maximising flexibility and hence effectiveness of water chilling machines in existing installations are examined - maximising cooling capacity and coefficient of performance (COP) of the machines themselves, and optimising the arrangements of machines in their water circuits for maximum cooling capacity and COP. Each of these methods is illustrated by a simulation study.

OPTIMIsING PERFORMANCE OF machiNes in existing WATER CHILLING installations

When seeking to optimise performance of machines in existing installations, the three key objectives, in order of priority, are to maximise, throughout the range of anticipated duties:

• refrigerating capacity,

• machine availability,

• energy economy in generating the required cooling.

The following guidelines are useful in achieving these objectives. For individual machines:

1. maximise refrigerating capacity by maintaining the minimum allowable refrigerant evaporating temperature, rather than just a specified or minimum water temperature. As it will be seen in simulation study A following, this is particularly beneficial in minimising the adverse effects of water-side fouling in the evaporators;

2. maximise COP by operating at maximum compressor efficiency. For most machines, this is likely to be at full compressor capacity.

In arranging machines in the water circuits of their installation:

3. if two or more streams of water at significantly different temperatures must be chilled, do not combine these streams before chilling. Rather use one or more machines to pre-chill the higher-temperature stream to an intermediate temperature approximating that of the other stream, as illustrated in Figure 1. Then combine the streams and use the other machines to chill the combined stream to the final required temperature. The advantage of so doing is that the refrigerating capacities of the machines pre-chilling the higher-temperature stream are maximised. This is because their refrigerant evaporating temperatures will be higher, and so if the condensing temperature stays approximately constant, the refrigerating effect per unit volume of refrigerant vapour drawn from the evaporator by the compressor will be higher, as shown for example by Gosney (1982);

4. arrange enough machines in parallel so that all the refrigerating capacity of all the machines is available to maximise chilled water delivery.

These guidelines were applied in the two simulation studies following. As these studies required detailed modelling of the internal behaviour of water chilling machines, they were conducted using the CHILLER computer program, which is now briefly described.

Figure 1. Chilling two water streams at different temperatures so as to maximise refrigerating capacity

the chiller computer program

This program, developed by the Chamber of Mines of South Africa, can predict the performance of complete water chilling installations, consisting of chilled and condenser water circuits incorporating conventional water chilling machines, cooling towers and water reservoirs interconnected in any user-specified configurations (Bailey-McEwan and Penman, 1987; Bailey-McEwan, 1991). CHILLER can simulate only conventional machines, which are larger versions of the integral, packaged design of water chillers developed for air-conditioning applications. Such machines employ a single- or multi-stage centrifugal compressor, a flooded shell-and-tube evaporator and a shell-and-tube condenser. In both these heat exchangers, the refrigerant passes through the shell and the water through the tubes.

The type of conventional water chilling machine in both simulation studies following is illustrated in Figure 2. A two-stage centrifugal compressor is employed, and the two corresponding stages of expansion are implemented by two expansion valves. An economiser is employed to collect the flash vapour generated in the first expansion valve; this vapour is then directed into the side inlet of the second stage of the compressor. The hot gas bypass valve opens at low part-duties when one or both stages of the centrifugal compressor would otherwise surge. The compressor has capacity-regulating inlet guide vanes on either its first stage or both stages.

Figure 2. Conventional water chilling machine with two-stage centrifugal compressor and economiser

CHILLER separately models each stage of a two-stage compressor, fundamentally modelling the interactions between stages and using artificial, empirical compressor curves for each stage (Bailey-McEwan and Penman, 1987; Bailey-McEwan, 1998). The principal features of these artificial curves are specifiable in order to match the curves of each actual stage as closely as possible. CHILLER can thus model the internal behaviour of such machines, as well as the detailed behaviour of both the chilled and condenser water circuits of an installation. It was thus suitable for the simulation studies now following.

simulation Study A: underground water chilling installation

This involved four conventional machines, employing Refrigerant 12, in an underground water chilling installation on a South African gold mine. Their specifications are given in Table 1. Their evaporators were connected in parallel in the chilled water circuit, as shown in Figure 3. Their condensers were likewise connected in parallel in the condenser water circuit, which rejected heat into the upcast air in an underground cooling tower. The two-stage centrifugal compressors had capacity-regulating guide vanes on their first stages only, and were controlled to maintain a chilled water temperature of 5°C.

No compressor characteristic performance curves were available from the manufacturer. Therefore, those artificial curves of CHILLER which best matched the limited amount of design, full-duty performance data furnished by the manufacturer were used.

When CHILLER was initially used to assess the performance of these machines, an unexpected prediction was that cleaning heat exchanger tubes to maintain the water-side fouling factors at design values was unlikely to save power! The reason is shown in the characteristic curves of the compressor's regulated first stage and unregulated second stage in Figure 4. Here, the operating points of these two stages are shown for the probable fouling factors (P), corresponding to the actual inlet guide vane opening of 50% on the first stage - and the lower, design fouling factors (D), at which the vane opening was predicted to be lower at 28%.

Table 1. Simulation study A: design, full-duty specifications of water chilling machines.

	Unit	Value
Evaporator
Water flow-rate	l/s	60,6
Inlet water temperature	°C	18,9
Outlet water temperature	°C	5,0
Water chilling load	kW(R)	3 520
Water-side fouling factor	m²°C/W	0,000176
Refrigerant temperature.	°C	3,2
Compressor
Number of stages		2
Mech. power absorbed	kW	1 024
Condenser
Water flow-rate	l/s	126,2
Inlet water temperature	°C	40,6
Outlet water temperature	°C	49,1
Water-side fouling factor	m²°C/W	0,000352
Outlet refrigerant temp.	°C	53,3
COP		3,43

Figure 3. Simulation study A: four machines in parallel in an underground installation

Figure 4. Simulation study A: effects on compressor of reducing water-side fouling factors in heat exchangers

When the water-side fouling factors of the heat exchangers are reduced to their design values by cleaning the tubes, the pressure rise, and hence the isentropic head, that the compressor must develop to maintain the required chilled water temperature reduces considerably. Because this compressor has guide vanes on its first stage only, the regulated first stage has to accomplish all this reduction in head. As seen in Figure 4, at the new vane opening of 28% at which this was accomplished in CHILLER's simulations, the curve of isentropic efficiency of the first stage was in a far lower range, and so its isentropic efficiency decreased sharply to 33 per cent. This was because the artificial compressor stage curves of CHILLER, used to model the real compressors, yielded maximum efficiency at the full-duty design point, and reduced efficiency elsewhere.

The real compressors were also likely to behave in this way, because they had been manufactured before 1980, when part-load performance began to significantly influence design criteria of conventional machines (Austin, 1991). Pre-1980 conventional machines, according to Austin's field monitoring experience, are most efficient at full load, which must mean that their centrifugal compressors have this characteristic. Thus when the heat exchanger tubes were cleaned, the decrease in input power owing to the reductions in refrigerant mass flow-rate and isentropic head was likely to be mostly or completely nullified by the increase in input power due to decreasing isentropic efficiency as the inlet guide vanes of the regulated first stage closed.

In sum, if such a machine is controlled to maintain a set outlet chilled water temperature - that is, if it operates under a “set water temperature” control philosophy - but is already on considerable part-duty, the vane opening of its regulated first compressor stage will be considerably below 100 per cent. This stage will therefore be close to or in the region of sharply decreasing isentropic efficiency. Power might thus not be saved when heat exchanger tubes are cleaned. Even more importantly, much of the machine's refrigerating capacity is unused and in fact rendered unavailable.

Multi-stage centrifugal compressors with only their first stages regulated are conveniently termed “first-stage-regulated” compressors. With such compressors, therefore, a preferable control philosophy is one which avoids low vane openings. The only way to do this is to increase the load. Therefore, there is advantage in cleaning the tubes and taking other measures to reduce compressor head if, by guideline (i) above, the “set water temperature” control philosophy is abandoned in favour of maximising the load on such machines, thus keeping the compressor vanes maximally open. This is a “maximum-load” control philosophy.

Table 2 gives the predicted performance of one machine in Figure 3 with design water flows and water-side fouling factors, but with actual inlet water temperatures. As seen, the predicted water chilling load of 2 430 kW(R) is 31 per cent below the design value of 3 520 kW(R) in Table 1. This is because the compressor vanes were only 38 per cent open, due to the considerably lower-than-design inlet water temperatures compared to Table 1!

There are three constraints on maximum loading of conventional machines: those imposed by the rated output of the compressor driving motor; by the surge lines of each compressor stage; and by the minimum allowable evaporating temperature. CHILLER was accordingly used to predict this machine's performance under the alternative “maximum-load” control philosophy, subject to these constraints, for otherwise identical conditions (Bailey-McEwan, 1998). Here, the compressor vane opening could increase to 70% before the evaporating temperature decreased to the minimum allowable value of 1°C. The resulting benefits of this “maximum-load” control philosophy were significant. Outlet chilled water temperature decreased to 2,8°C, water chilling load increased to 3 049 kW(R), and COP increased to 3,42. Compared to Table 2, the improvements in water chilling load and COP were 25 and 15 per cent respectively! The lower-than-design chilled water temperature could only be beneficial, as these machines were serving air cooling cars.

Table 2. Simulation study A: predicted performance of one machine with actual inlet water temperatures.

	Unit	Value
Evaporator
Water flow-rate	l/s	60,6
Inlet water temperature	°C	14,85
Outlet water temperature	°C	5,27
Water chilling load	kW(R)	2 430
Water-side fouling factor	m²°C/W	0,000176
Refrigerant pressure	kPaa	347
Refrigerant temperature	°C	3,63
Economiser
Refrigerant pressure	kPaa	574
Refrigerant temperature	°C	20,40
Condenser
Water flow-rate	l/s	123,1
Inlet water temperature	°C	34,78
Outlet water temperature	°C	41,08
Water heating load	kW(R)	3 246
Water-side fouling factor	m²°C/W	0,000352
Refrigerant pressure	kPaa	1 063
Outlet refrig. temperature	°C	44,19
Compressor
Vane opening	%	38
1st stage inlet vol. flow	m³/s	0,90
1st stage isentropic head	kJ/kg	8,78
1st stage isentr. efficiency	%	37,4
2nd stage inlet vol. flow	m³/s	0,73
2nd stage isentropic head	kJ/kg	12,08
2nd stage isentr. efficiency	%	67,6
Input power	kW	816
Suction temperature	°C	3,63
Discharge temperature	°C	73,90
COP		2,98

Compressor vane opening was still significantly below 100 per cent, though, so a further idea was considered to improve machine utilisation. As shown in Figure 3, there were four such machines in the installation. Performance was next predicted, therefore, with the evaporator water flow-rate increased from 60,6 l/s to 80 l/s (where the water velocity through the tubes just reached the recommended maximum of 2,2 m/s). The total required chilled water flow-rate of 240 l/s could then be delivered by three machines, freeing one machine for maintenance during normal working hours. For the other conditions remaining as before - except that condenser water flow-rate was slightly increased to the design value in Table 1 - compressor vane opening could now increase to 97 per cent, being limited there by the refrigerant evaporating temperature again reaching 1°C. Predicted outlet chilled water temperature, water chilling load and COP were 3,47°C, 3 812 kW(R) and 3,49 respectively. Again compared to Table 2, the improvements in water chilling load and COP were 46 and 14 per cent respectively! Here, the maximum-load control philosophy thus offered improved machine availability (through freeing one machine for maintenance), as well as greatly improved effectiveness and significantly improved quality of performance.

Of course, the accuracy of these quantitative predictions was entirely dependent upon the correspondence between the artificial compressor stage curves of CHILLER and those of the real machine. Nevertheless, the qualitative benefits of this alternative, “maximum-load” control philosophy for conventional machines are clear. First, the machine always accepts the maximum water chilling load possible for the prevailing operating conditions. Other things being equal, this means, as in this study, that if heat exchanger tubes are cleaned, the benefit of lower chilled water temperature is immediately obtained. A “set water temperature” control philosophy yields no such benefit! Alternatively, if it is desirable to increase chilled water flow-rate, the machine will chill this new flow to the maximum possible extent. Second, for pre-1980 conventional machines with first-stage-regulated compressors, as in this study, the optimal machine COP is likely to be attained, because the compressor inlet guide vanes are maximally open, hence maximising the efficiency of the regulated first stage. Optimal performance in two aspects - water chilling capacity and COP - results.

simulation Study b: surface water chilling installation

This involved a combined bulk air cooling and service water chilling installation, as illustrated in Figure 5, situated on surface at a South African gold mine. The three-cell bulk air cooler (BAC) processed 760 m³/s of ambient air proceeding underground, requiring 450 l/s of chilled water at 3°C. The shaft's anticipated requirement for chilled service water was 20 Ml per 24 hours. As is customary in South African mine cooling practice, this service water was obtained from the water pumped out of the mine, which reached surface at approximately 27°C throughout the year. As seen in Figure 5, a five-cell pre-cooling tower was thus used to perform the first step of chilling the service water. The design water flow-rate through this tower was 265 l/s - slightly exceeding the average flow-rate of 252 l/s necessary to deliver 20 Ml of service water over 24 hours.

Figure 5. Simulation study B: bulk air cooling and service water chilling installation

The task of the arrangement of water chilling machines in Figure 5 was to chill both the pre-cooled service water and the water leaving the BAC to 3°C. It was proposed to make use of four existing, conventional water chilling machines - used in an earlier version of the installation - in a twin series-parallel (SP) arrangement, shown in Figure 6. Machines 1 and 3 there, performing the first step of chilling, are termed the lead machines, and the others are termed the lag machines. Their specifications are given in Table 3. Unlike simulation study A, the compressors had capacity-regulating guide vanes on both stages.

In the proposed SP arrangement of Figure 6, the water from the BAC was fed into the cool dam also receiving the pre-cooled service water; the combined flow of water to be chilled was then evenly split between the two parallel branches of lead and lag machines. As also shown, automatic flow-control valves were installed in these branches. The mine required one machine to be shut down during the periods of peak electrical demand, 09:00-11:00 and 13:00-15:00, termed the peak demand periods. During these periods, it was intended that the corresponding flow-control valve reduces the water flow in its branch to enable the other, operating machine to deliver water at 3°C.

The condensers of the machines were all connected in parallel in their water circuit, which rejected heat to the atmosphere in a four-cell heat rejection cooling tower.

Figure 6. Simulation study B: proposed series-parallel (SP) arrangement of water chilling machines

Table 3. Simulation study B: specifications of water chilling machines.

	Unit	Machines
				Lead	Lag
Evaporator
Water flow-rate	l/s	350	350
Inlet water temperature	°C	17,33	9,45
Outlet water temp.	°C	9,45	3,00
Water chilling load	kW(R)	11 550	9 450
Refrigerant temperature	°C	6,42	0,52
Compressor
Number of stages		2	2
Mech. power absorbed	kW	1 755	1 755
Condenser
Water flow-rate	l/s	670	670
Inlet water temperature	°C	27,50	27,50
Outlet water temp.	°C	32,24	31,50
Outlet refrig. temp.	°C	36,12	34,78
COP		6,58	5,38

At first sight, the SP arrangement of Figure 6 seemed thermodynamically well-designed and practicable, with simple piping requirements. The machines chilled water in two steps, so the lead machines had higher refrigerating capacity and COP. CHILLER was used to predict the performance of the whole installation with this arrangement, with the BAC operating continuously. The lead machines were kept at full capacity, and the compressor inlet guide vanes of the lag machines were regulated to deliver water at 3°C. Figure 7 shows the vane openings of the lag machines and the predicted average hourly delivery of chilled service water at 3°C in mid-summer, when the average ambient wet-bulb temperature varied between 19,4°C and 15,7°C. This delivery was the maximum of 265 l/s from all five cells of the pre-cooling tower, except during the peak demand periods, when Machine 4 was shut down. Then, only three cells of the pre-cooling tower operated and the delivery dropped to an average of 111 l/s, the flow through Machine 3 being reduced so that it could deliver water at 3°C. Over 24 hours, 20,7 Ml of chilled service water was delivered, meeting the anticipated requirement of 20 Ml.

Figure 7. Simulation study B: service water delivery of SP machine arrangement in mid-summer

In this study, the maximum flow through each machine's evaporator was limited to the design value of 350 l/s in Table 3. This was because the design refrigerant evaporating temperature, also in Table 3, of the lag machines was low at 0,52°C. Greater-than-design flows through these machines thus incurred the risk of the evaporating refrigerant temperature falling below 0°C, and so of water freezing in tubes. In fact, at the maximum vane opening of 84 per cent in Figure 7, the predicted evaporating temperature was 0,43°C. Unfortunately, therefore, the maximum service water delivery rate of the SP arrangement was limited to the mid-summer value of 265 l/s, because the water flow through each machine - 357 l/s in Figure 6 - corresponded to the design value of 350 l/s. In turn, this meant that some of the refrigerating capacity of the lag machines was unused and unavailable - as seen from the vane openings in Figure 7, which were far short of 100%. Moreover, during the rest of the year, even though the pre-cooling tower was more effective and the load on the BAC was less, this arrangement could deliver no more service water - except in winter, when the BAC was shut down.

Furthermore, this SP arrangement did not maximise the capacity of the water chilling machines, for the following reason. In mid-summer, the BAC, designed to accept water at 3°C, returned its 450 l/s of water at an average temperature of 12,4°C - 7,5 degrees lower than the average temperature, 19,8°C, of the 265 l/s of water leaving the pre-cooling tower during all but peak demand periods. Thus the two water streams to be chilled were at considerably different temperatures, and refrigerating capacity is lost if these streams are combined before chilling - as they are in Figure 6.

Guidelines (iii) and (iv) above were therefore used to suggest an alternative arrangement: the junction series-parallel (JSP) arrangement of Figure 8. As the flow through the BAC, 450 l/s, was far higher than the average service water demand of 265 l/s, it was logical to use one machine to pre-chill the pre-cooled service water. This pre-chilled service water could then be merged with the returning water from the BAC, and the other three machines in parallel could then perform the final step of chilling, as shown in Figure 8.

Figure 8. Simulation study B: alternative, junction series-parallel (JSP) arrangement of machines

The advantage of this arrangement was twofold. By guideline (iii) above, because Machine 1 received pre-cooled service water at a relatively high temperature, its refrigerating capacity was higher than those of the lead machines in Figure 6. Second and more importantly, by guideline (iv), arranging the other three machines in parallel enabled all the capacity of the machines to be utilised to increase service water delivery if required. Even when this delivery was increased to the maximum at which all four machines, operating at full capacity, could chill water to 3°C, the refrigerant evaporating temperatures of Machines 2 through 4 were above their design value of 0,52°C in Table 3. This was because the flow-rate through each of these machines was always below the design value of 350 l/s, and their inlet water temperatures were always higher than the design value of 9,45°C in Table 3. Accordingly, this JSP arrangement was simulated to maximise the delivery rate of chilled service water. When this exceeded 350 l/s, the flow control valve in Figure 8 opened to bypass the excess round Machine 1. Unlike the SP arrangement, all four machines could operate at full capacity in mid-summer, except during peak demand periods, when one machine had to be shut down. In spring/autumn, when three machines were sufficient, Machine 1 was shut down and Machines 2-4 were operated, still in parallel to maximise delivery of service water.

Figure 9 shows how much extra chilled service water the JSP arrangement delivered in mid-summer. As the hourly wet-bulb temperature decreased, so the delivery rate could increase, exceeding 400 l/s during the coolest part of the night. Over 24 hours; the JSP arrangement was predicted to deliver 27,7 Ml of service water, 34 per cent more than the 20,7 Ml delivered by the SP arrangement. Moreover, the predicted overall 24-hour COPs of the two arrangements were almost identical at 6,34 and 6,33 respectively. Even more striking was the comparative delivery in mid-spring/mid-autumn, when the average wet-bulb temperature varied between 9°C and 13,8°C. Here, the JSP arrangement delivered 31,6 Ml - 53 per cent more than for the SP arrangement, which remained limited to 20,7 Ml for the reasons given above. Clearly, therefore, the JSP arrangement offered far more effective utilisation of the same capital equipment.

Figure 9. Simulation study B: service water delivery
of SP and JSP machine arrangements in mid-summer

conclusion

The two simulation studies have proposed ways of making more effective use of capital equipment - the expensive water chilling machines in mine cooling systems. Implementing such proposals in any particular installation requires the close co-operation of the machine manufacturer. Here, in the first study, implementing the proposed “maximum-load” control philosophy requires the manufacturer's accurate knowledge of the compressor characteristics, in order to prevent surging and overloading of the compressor driving motor. In the second, maximising service water delivery amounts to another “maximum-load” philosophy, and the machine controllers must moreover keep the refrigerant evaporating temperatures safely above 0°C. (Service water delivery in the simulations for mid-spring/mid-autumn was slightly limited because of this.) Also, in the second study, the flows through Machines 2 through 4 are considerably below their design values, so measures would have to be taken to prevent suspended solids from settling and fouling evaporator tubes. Nevertheless, the value to a mine of maximising the cooling capacity of its existing installations may well justify the expense of such modified controllers and other measures to make new operating policies practicable.

references

Austin, S.A., 1991, “Optimum Chiller Loading,” ASHRAE Journal, Vol. 33, No. 7, July, pp. 40-43

Bailey-McEwan, M. and Penman, J.C., 1987, “An Interactive Computer Program for Simulating the Performance of Water Chilling Installations on Mines,” Proceedings, APCOM 87: 20th International Symposium on the Application of Computers and Mathematics in the Mineral Industries, Vol. 1: Mining, Johannesburg, South African Institute of Mining and Metallurgy, pp. 291-305

Bailey-McEwan, M., 1991, “Use of the CHILLER Computer Program with Conventional Water Chilling Installations on South African Gold Mines,” Journal of the Mine Ventilation Society of South Africa, Vol. 44, No. 1, January, pp. 2-12

Bailey-McEwan, M., 1998, “Assessing Performance of Large Mine Water Chilling Machines Using Refrigerant-Circuit Measurements and Machine Modelling,” PhD Thesis, University of the Witwatersrand, Johannesburg, South Africa

Chamber of Mines of South Africa, 1991, private communication

Gosney, W.B., 1982, Principles of refrigeration, Cambridge University Press, pp. 203, 208, 209.

van der Walt, J., 1979, “Engineering of Refrigeration Installations for Cooling Mines,” The South African Mechanical Engineer, Vol. 29, No. 10, October, pp. 360-372

To avoid freezing in tubes, this must be greater than 0°C, but it may have to be still higher to avoid surge in one or more compressor stages if the design evaporating temperature of a machine is well above 0°C.

The reasons for this were complex; in sum, in the JSP arrangement, Machine 1 was predicted to operate less efficiently, and Machines 2 through 4 more efficiently, than in the SP arrangement.

2

I SZKOŁA AEROLOGII GÓRNICZEJ 1999

5

776

PROCEEDINGS OF THE 7^TH INTERNATIONAL MINE VENTILATION CONGRESS

775

THE ROLE OF WATER CHILLING MACHINES IN OPTIMISING PERFORMANCE

---