Table of Contents
Solution mining followed by solvent extraction-electrowinning (SX-EW) is one possible way to economically extract copper from ore bodies which cannot otherwise be profitably exploited. Published costs for the unit processes of SX and EW are out of date especially in view of the newer reagents now available and the newer circuit configurations. This paper updates the costs for SX-EW and also discusses some of the trade-offs which must be made when designing an SX-EW plant.
The pairing of the unit operations solvent extraction-electrowinning (SX-EW) is an accepted technology for the recovery of high quality copper and it has been used in combination with in situ leaching for copper recovery. The object of this paper is to present the capital and operating costs in 1985 U.S. dollars and other units for the unit operations of SX-EW only so that those companies contemplating the use of this technology for final copper recovery from in situ or solution mining leach liquors will have a good starting point for preliminary cost estimations. Some detailed costs for individual plants have been published but in most cases the information is not current. In addition, a complete survey of all operating plants has not been done.
Presentation of the Data
The data on capital costs was made available from Bechtel which has experienced specialists in SX-EW design and construction. It is important for the reader to realize that, while these data were derived from actual case studies and are thus useful for preliminary cost estimations, each situation is unique and accurate capital costs for any new plant to be built can only be obtained via a comprehensive feasibility study.
The data on operating costs were taken from a detailed questionnaire which was sent to all operating copper SX-EW plants. Responses from all but two of the plants were received. For some plants the data represent several or more years of operation while for more recent plants, some still in the startup phase, the data are necessarily not representative of the long term. For all plants the information provided is interesting, informative and most appreciated.
Many of the costs are presented in generic terms, i.e., units per kilogram of copper produced, tonnes of copper produced per man year, etc. This was done so that:
- The operating costs in monetary units for the responding companies are not revealed.
- Companies which are contemplating the use of SX-EW can calculate what their monetary costs might be by taking the costs in generic terms and factoring in local conditions i.e., power costs, labor costs, etc.
Solvent extraction (SX) and electrowinning (EW), though chemically interdependent in an SX-EW plant, are two very distinct and physically separate unit processes.
There are things that can be done in the SX circuit which will alter its capital cost but will not change the tankhouse and vice versa. For these reasons this paper will discuss the capital costs for each unit process before combining them into a total unit cost.
Items that make up the capital cost of an SX circuit include: 1) General site development and utility installation, 2) Pregnant leach solution (PLS) ponds and pumps, 3) Raffinate pond, 4) Mixer-Settlers, 5) Organic tanks and pumps, 6) Crud treatment system, and 7) Initial fill of diluent and copper extractant. For normal dump leach systems the PLS and raffinate ponds are much more costly items than they would be for an in situ leach operation.
The capital costs in 1985 U.S. dollars for some recent case studies are shown in Table 1. In general the capital cost for the SX circuit is directly related to the volume of PLS to be treated and closely tied to the staging of the circuit. This is expected since the size of each of the items, except for reagent concentration, is dependent on the total solution flow.
The 2 extraction-1 strip stage configuration for Cases 1, 4 and 6 through 9 is typical for several recently completed plants while Cases 2, 3, and 5 represent examples of further reduced staging. From the data it is obvious that by reducing the staging the unit capital cost, based on flow, can be lowered significantly (Case 2 and 3) or at least be kept competitive even when the circuit must be enclosed in a building (Case 5 ).
Unit cost data based upon flow are also shown in Table 1. If the 2 extraction-1 strip stage circuits only are considered it is obvious that the unit capital costs are higher for the lower flow ranges. Further analysis of the data shows that there is a levelling off at an average of about $6600/m³ /h for a PLS flow of 1800 m³/h and above.
Unit capital costs based on an annual tonne of copper produced are also shown in Table I. The general trend is that the greater the copper content of the leach liquor the greater the copper transfer per unit volume and thus the lower the capital cost per tonne of annual copper production.
Specific items which make up the capital cost of the electrowinning tankhouse include: 1) Tankfarm tanks and pumps, 2) Electrolyte filtration and heat exchanger installations, 3) Tankhouse building including corrosion resistant flooring and ventilation systems, 4) EW cells and support systems 5) Anodes and cathodes along with their support and spacing systems, 6) D.C. electrical system including rectifiers and bus, 7) Cathode washing and handling equipment, 8) Piping systems, and 9) Safety systems.
Table 2 presents Summary Electrowinning Capital Cost Data for the cases previously discussed. As expected, the capital cost for EW is a function of the tonnage of cathode produced in that the more copper to be produced the higher the capital cost. It is also not surprising that in general, the unit capital cost decreases as the production capacity increases. What may look to be a little surprising is the wide range in unit capital costs, based on copper produced, from a high of $1390/annual tonne to a low of $710/annual tonne. In fact Case 4 ($1390/annual tonne Cu) is for an unusual situation because of climatic conditions and because it was designed to produce its own copper starter sheets. The “normal” range then is $710 to $1210/Ann tonne of copper produced, which is still quite broad. However, this broad range is not surprising when one considers the major factors which impact the unit capital cost of electrowinning: 1) Degree of automation in starter sheet preparation and in cathode handling, 2) Type of cathode starter sheet system, i.e., copper starter sheets or stainless steel blanks, 3) Whether starter sheets are produced or purchased, 4) Current density and expected efficiency, 5) Degree of weather proofing for the tankhouse, 6) Project life, i.e., quality and type of materials of construction.
Total SX-EW Capital Costs
Table 3 summarizes the total combined SX and EW capital cost for the case studies discussed previously. As with the individual SX and EW capital costs the total SX-EW unit capital cost tends to decrease with increasing copper production capacity.
It is fairly easy to separate the capital cost for SX from that for EW. For this reason one might choose to categorize the operating costs in the same way. However, because of the way SX-EW plants are designed and operated and the manner in which costs are allocated this can be a difficult task. The approach of this paper is to treat labor and power requirements for SX-EW as a whole but to assign most other costs to either SX or EW.
Table 4 shows the labor requirements of the 13 operating plants that reported them. Note that many of the plants reported total labor but did not break it down into the SX and EW operations. This is not surprising since many plants, especially small ones, expect some of the employees to work in both areas. In addition, some of the plants included their maintenance employees while others did not. At times the information received was not clear and if it has been interpreted incorrectly for some plants we apologize.
A very interesting number to look at is the annual copper production per employee since this number represents a total labor requirement per unit of production i.e., productivity. The calculated value runs from a high of 550 tonnes Cu per employee to a low of 158. Reported values in many cases did not agree with calculated values and we have no explanation for this except in some instances where the total employees, including buying, administration, etc. may be included in the productivity figure.
In general, plants with larger stages and/or higher copper contents in the leach solution have the highest SX productivity. This is to be expected since the control points on the SX unit operation are the same regardless of its size, thus more copper can be transferred per employee when flows per stage are larger or the copper content of the leach solution is higher. Labor requirements for EW can also vary depending on the degree of sophistication of the tankhouse, the size of the copper pulls, and whether starter sheets have to be plated, pulled and prepared or are purchased. Considering this the productivity for plant N is to be commended since it produces its own starter sheets, has only a small hand controlled crane for pulling copper and treats a leach solution with a relatively low copper concentration.
Several new SX-EW plants, whose data is not reported here, have copper productivity values about twice as high as the best reported in this paper. These plants treat large flows with minimum staging and do not produce their own starter sheets. In one case the tankhouse is highly automated with cathode being plated on stainless steel blanks and automatically stripped.
When SX-EW technology was first introduced two of the major concerns of the copper producers were the possibility of high reagent losses and reagent contamination. The authors are not aware of any case where reagent contamination has been a severe problem in the type of SX plants under discussion. This is due primarily to two reasons: a) The operating companies practice good circuit hygiene. They generally do not add anything to the circuit without first talking to reagent suppliers and if needed, either run themselves or have the reagent supplier run, compatibility tests; b) many contaminants wash out of a circuit with time.
Generally reagent losses in most operating circuits are not excessive. Table 5 shows reagent usage data for many of today’s operating plants along with specific flow per unit of settler area. Of particular interest is the column showing reagent consumption per unit of copper produced. Note that this number for steady state operating plants ranges from a high of .0078 Kg reagent/Kg Cu to a low of .0015 Kg reagent/Kg Cu with an average of .0046 Kg reagent/Kg Cu.
Reagent losses can be classified as physical or chemical. Physical losses include:
a) Entrainment of organic in exiting streams (raffinate and electrolyte)
b) Loss of organic in circuit crud
c) Leaks and/or spills
Entrainment losses are usually caused by very small organic droplets in the aqueous phase which do not have a chance to coalesce into larger drops and float out of the aqueous phase. Factors which should decrease entrainment are: 1) longer settler residence time, 2) proper emulsion distribution so that even flow down the settler is obtained, 3) proper mixing, 4) the correct phase continuity, 5) avoiding a high viscosity in the feed solution, 6) steady state flows, and 7) higher staging. This discussion assumes the settlers are of sufficient size so that emulsion is not passing beneath the aqueous underflow-weir. If this happens organic losses become excessive and immediate action is necessary.
Losses of organic to circuit crud can be significant. For example, plant B, which has a relatively high reagent usage, treats an agitation leach liquor and crud formation in the plant is higher than normal. Most plants try to recover circuit organic from the crud by treating the crud in a centrifuge. This is now a standard piece of equipment for new plants. Further evidence for the effect of crud formation on reagent usage comes from a recent report. The plant of Burro Chief Copper Company at Tyrone, New Mexico had been operating for about 18 months using the tridecanol modified aldoxime reagent LIX 622. In November of 1985, tankhouse capacity was doubled and a substantial quantity of the unmodified ketoxime reagent LIX 84 was added to the SX circuit so that the copper transfer of the circuit was nearly doubled. Immediately crud began to leave the circuit in the raffinate and for the past six months the plant has not had to run their crud cleanup system.
Over that time period copper production has almost doubled while total reagent usage has dropped substantially. It should be emphasized that the phenomenon seen in the Burro Chief Copper Company SX circuit is not fully understood and that each SX circuit is unique. Thus, what is seen in one plant may not be seen in another. However, the events in the Burro Chief circuit are consistent with other observations on the role of modifiers in crud formation in both pilot plants and one commercial circuit.
Organic losses by leaks and/or spills should be insignificant in a welt operated and maintained plant.
The major chemical loss of reagent is by hydrolytic degradation. This topic has been discussed by several authors and there is some disagreement on how serious the problem is. Analyses of circuit organics show that in most plants running at steady state the concentration of degradation product reaches a certain level and then holds fairly constant. At this point the rate of formation of degradation product is equal to its rate of loss. Assuming that the majority of the degradation product is lost via entrainment and knowing reagent usage in the plant the amount of reagent lost by degradation can be calculated . These calculations show that degradation losses are .001 Kg reagent/Kg Cu or greater for some plants but much less for others. Factors which result in increased reagent degradation are higher acid concentration and higher temperatures in stripping.
The structure of the reagent is also important. Laboratory experiments suggest that:
- Ketoxime extractants are more stable than aldoxime extractants.
- Aldoxime extractants based on 5-dodecylsalicylaldoxime are more stable than those based on 5-nonylsalicylaldoxime.
- The use of equilibrium modifiers with aldoxime extractants enhances the rate of reagent degradation.
In looking at the data in Tables 5, 10, and 11, it is very difficult to find trends and draw any conclusions about the relationship of reagent usage to specific flow in the settler, tankhouse acid, and staging. Reagent usage in plants B, D, and M appear to be higher than normal while that for plants A, C, E, F, and N are lower than normal when staging, tankhouse acid and settler size are considered.
The selection of a proper kerosene is important since the organic phase in the copper SX plants under discussion is 80% or more diluent. Kerosene usage is a function of the entrainment of organic phase in the exiting stages with some diluent loss also due to evaporation. Of interest is the fact that several operating plants do not cover their settlers (D, G, I, L, and N). It might be expected that this would result in a significantly higher diluent use because of evaporation. Table 5 shows the volume % of reagent reported in each of the operating circuits and the volume % of reagent added along with kerosene for the past year or more. In most plants the circuit is operating at a slightly higher reagent concentration than is added as makeup. This is consistent with the circuits losing a small amount of diluent by evaporation. In reality the loss of diluent by evaporation is much greater. Since reagent is lost from the circuit by degradation in addition to entrainment, the best way to determine diluent loss through evaporation would be to compare the reagent concentration added as makeup to the total concentration of reagent and degradation product in the circuit. Interestingly, there is one plant (J) where the reagent makeup is added at a much higher concentration than the circuit organic. If the concentrations of active reagent and degradation product in this circuit organic are added the sum compares closely to the concentration of reagent added as makeup, which indicates there is little or no loss of diluent by evaporation. This plant has tightly covered mixers and settlers thus it looks like tight mixer-settler covers may reduce kerosene consumption.
In several plants reagent makeup is at a much lower concentration than circuit organic. This indicates that the reagent concentration was decreasing for the time period reported.
Major Tankhouse Costs (Table 6)
Anode Consumption. The data received on anode consumption was not as complete as much of the other data but it does show a wide range of anode usage; from a high of .171 anodes/tonne Cu to a low of 0.013 anodes/tonne Cu with an average of 0.067 anodes/tonne Cu. The data suggests that Pb-Sb anodes are consumed faster than Pb-Ca but this conclusion is tenuous since it is based on a very small sample. A recent presentation suggests that the method of manufacture may have more to do with the anode stability than the alloy. Research on anodes is continuing and anode usage is expected to lessen as anode quality increases.
Electrolyte Bleed. The only way to control impurity levels (primarily iron) in the recirculating electrolyte stream is to bleed the electrolyte at whatever rate is required to hold the impurity at the desired concentration. With the excellent copper over iron selectivity displayed by today’s reagents tankhouse bleeds are normally less than 5% of the electrolyte advance flow. Since the bleed reports either back to the extraction section or to the raffinate the copper is recycled and in most cases the acid is consumed in leaching. Cobalt in the electrolyte, used to reduce lead contamination in the cathodes, is lost but the cost is not a major one. In the rare case where the bleed is not recycled the costs associated with it are much more important.
Power Consumption. The electrowinning process is power intensive, thus, current efficiency is very important, especially where power costs are high or where a more efficient use of power yields increased copper production. Current efficiencies in the high 80 to low 90% range are normal and most EW plants surveyed operate in this range. High current efficiency and superior quality copper go hand in hand with a well run tankhouse. The small costs associated with running a tankhouse well are normally far less than the resulting savings in power and increased price for copper. Overall power consumption in EW should be a little less than 7.92 KJ/Kg Cu (1 Kwh/Lb Cu).
Miscellaneous Tankhouse Costs. Electrolyte filters to remove entrained organic, mist suppressants, smoothing agents and maintenance supplies all add to the EW costs. Normally these combined costs are not high, amounting to less than 3-4 U.S. cents/Kg Cu, and some plants run quite well without any of these items other than maintenance supplies.
Summary of Costs
Table 7 summarizes the major operating costs for those SX-EW plants which supplied data to us. As is obvious from both the capital cost data previously discussed and the operating cost data, there is no such thing as a typical or average SX-EW plant. Each situation is different and each plant has its own unique operating problems and successes. For these reasons a company considering SX-EW cannot use these data to predict their expected cost in an exact manner but they can use these data to predict what is reasonable within a certain range.
A case study is presented for the following reasons:
- To point out the modern trends in SX-EW operations.
- To discuss some of the choices facing companies contemplating the use of SX-EW.
- To show reasonable operating costs using the more recent plants as models.
Consider a situation in the United States where an in situ leach solution (PLS) containing 1.36 g/l Cu and 2.0 g/l Fe3+ at a pH of 1.86 is to be treated by SX-EW. The goals are to recover 90% or more of the copper from the leach solution while producing an electrolyte from which high quality copper can be electrowon, all at minimum total cost. PLS flow is 910 m³/h.
The trend today with typical dump leach solutions is to use circuits with 2 extraction stages and 1 strip stage (2×1 circuit) or in some cases further reduced staging consisting of 1 extraction stage coupled with 2 extraction stages and 1 stripping stage (1x2x1 circuit). Continuous circuit data generated in the laboratory is shown in Table 8 for the circuit configurations mentioned above. All of the objectives have been met.
The total capital cost for the two SX-EW plants under consideration can be approximated (Table 9). These costs are based on the following assumptions. For the 2×1 SX plant the capital cost was obtained by averaging the capital costs for the 3 SX plants with PLS flows of 1820 m³/h in Table 1, then dividing by 2 and multiplying by 1.1 since the costs at ½ scale should be about 10% greater on a unit flow basis. For the 1x2x1 circuit the PLS flow is split between the single extraction stage and the 2 counter current extraction stages. As a result each mixer settler is ½ the size of those in the 2×1 circuit and the total settler area is only 66.7% of that of the 2×1 circuit. The cost saving is estimated to be about 29% which gives a capital cost of $5.0 million. The electrowinning capital cost is the same for each plant and is based on a copper production of 27.2 tonnes/day. A unit capital cost of $945/Ann. tonne, which is about midway between the high and the low unit costs for the case studies shown in Table 2, was used. Actually, this cost is probably a little high for a standard tankhouse which will not produce its own starter sheets.
Operating costs are a little more difficult to predict than capital cost. The approach this paper takes is to calculate costs which are based on the operating data reported by the industry tempered with the authors experiences in supplying reagents and technical service to operating plants over the past 20 years. The predicted operating costs are shown in Table 9. These costs are given in generic units and also in cents/lb Cu with the following assumptions being made: total power includes EW plus about .3 kwh/lb Cu for SX with a cost of 7 cents/kwh; reagent losses are 100 ppm with a cost of $5.50/lb reagent; kerosene is $1.10/gallon; anodes are $230/anode; total employees are 18 including a ½ time middle manager and a ½ time secretary for a total cost of $625,000/year. Since in situ leach solutions will have good clarity and a constant moderate temperature SX circuits will run very smoothly. Under these conditions the predicted operating costs should be easily attainable if proper attention is given to the plant operation.
The data in Tables 8 and 9 gives a feel for one of the important decisions a company faces when deciding the type of SX circuit they should build. The company can spend a little extra to build a 2×1 plant and have lower operating costs or it can save a little on capital by building a 1x2x1 plant and pay higher operating costs. In our example $2 million in capital can be saved by going to a 1x2x1 circuit but the overall costs are higher by 2.38 cents/lb Cu ($521,220/year) . For the case under study with the assumptions that were made the 2×1 plant is the best choice.
The added operating cost of the 1x2x1 circuit is all due to organic losses. If a loss of 50 ppm instead of 100 ppm is assumed, an assumption very much in line with several recently completed plants, then the difference in operating cost is 1.19 cents/lb Cu ($260,610/year). Now the decision is a very difficult one. Certainly a well designed and operated SX plant can run with reagent losses of 50 ppm, however, because each plant is unique and each leach solution different it is very difficult to predict ahead of time exactly what the reagent losses will be.
Suppose that the copper content of the feed solution is about half of that in the case study so that copper production is about 13.6 tonne/day. In this case the increase in operating costs for the 1x2x1 versus 2×1 circuit would be $261,000 assuming 100 ppm losses and $130,500 assuming 50 ppm losses, respectively. For the latter instance the decision is much easier.
Another difficult decision facing companies is the choice of reagent. As mentioned earlier, there exists compelling evidence from both pilot plants and commercial plants that in some circuits the use of mixed ketoxime-aldoxime reagents results in significantly lower overall reagent usage. It is unfortunate that at the present time there are no simple laboratory tests with good predictive value in this area. A carefully and properly run pilot plant trial is more reliable but in some cases only a commercial plant test will provide the information needed so that the operator knows which reagent or reagent mixture is best in his plant.
SX-EW for copper recovery from leach solutions has proven to be a low cost technology for the production of high quality copper. However, it is evident from the data provided by the industry that some plants have much lower operating costs than others (Table 7). It is hoped that the individual operating plants will be able to use the operating data presented in this paper as a standard by which to measure their operation. Problem areas can then be identified and solutions to the problems sought. In the same way companies contemplating the use of SX-EW technology can make use of the experience of operating plants in their design, construction and operation plans. The best sources of information are the operating plants, those engineering companies which have specialists in SX-EW and reagent suppliers. The industry should take advantage of all of these sources of information as early in the planning stages as they can. The result should be a better designed, smoother operating plant.
Attached as an appendix are Tables 10 and 11 which give much of the design and typical metallurgical data for those operating SX-EW plants which answered the questionnaire. These data are given so that the readers can appreciate the variety of feeds being treated, the various plant designs in use and the different operational parameters.