Global Copper Production

To assess the risk earthquakes may pose to the global copper supply, information on copper facilities and their 2018 production data were collected. Construction of a global copper facilities dataset began by gathering lists of copper producers from multiple sources, including the USGS National Minerals Information Center (NMIC; Xun, 2020; Chung, 2021; Flanagan, 2021; Safirova, 2022; Soto-Viruet, 2022a, b; Yager, 2022), the International Copper Study Group (2015), CRU International Ltd. (2018), and S&P Global (2019). Facilities were cross-checked and combined to create an initial list. Names, present and past ownerships, capacities, and approximate locations were used to identify, verify, add, and eliminate entries. Each location was verified and geolocated using Google Earth (Google LLC., 2019) and stored in a geodatabase. Facilities were excluded from the dataset if it was not possible to identify their locations.

The amount of copper produced in 2018 was collected for each facility by production stage (mine, smelter, or refinery; fig. 1A, B, and C). Production data for each facility were determined hierarchically. Reported company data were considered as the first-tier source. If data were reported in a company report, the production quantity was checked against a facility’s capacity to confirm they were the same order of magnitude, and if so, the data were added to the dataset. If the reported production quantity varied from a facility’s capacity considerably, subsequent sources were used to confirm the production quantity. If none were found, the company-reported production quantity was assumed correct and added to the dataset. Government reporting was considered as the second-tier source; quantities were accepted following the same validation steps used for company data. When company and government data were unavailable, a third-tier reported production quantity that aligned closest to a facility’s capacity was selected. These third-tier sources consisted of news reports and subscription sources. For facilities that had no reported 2018-production data, production was estimated. Estimation methods included using reported production quantities for 2017, using a ratio of a facility’s production capacity to total country production or to reported company production, or calculating production from mined ore production when the copper content was known. If estimating production using one of the described methods was not possible, then facilities were excluded from the dataset.

Although the USGS NMIC surveys domestic producers for monthly and annual copper production, data are aggregated or withheld from publication to avoid disclosing proprietary information. For that reason, proprietary data reported to the USGS NMIC were not included in the dataset. Production data for U.S. facilities were collected using the same open and subscription sources and estimation methods that were used for international producers. Overall, a total of 625 copper-producing facilities that had 2018 production data were identified and validated using these methods. The excluded facilities are discussed in the “Geospatial Concentration of Copper Production” section.

Geospatial Concentration of Copper Production

In 2018, world copper mine production estimated by the USGS was about 20.4 million metric tons (Mt); Chile and Peru were the top producers (Flanagan, 2021). Because data collected by facility can be limited by a lack of available production or location data, world mine production captured for this analysis was 18.7 Mt (about 92 percent of world 2018 production reported by the USGS). The USGS reported that 2018 world copper smelter production was 19.5 Mt and world copper refinery production was 24.4 Mt (Flanagan, 2021). Smelter production data for this study totaled 16.2 Mt (about 83 percent of the USGS 2018 total) and refinery production totaled 20.9 Mt (about 86 percent of the USGS 2018 total; table 1).

Chile (5.6 Mt) and Peru (2.3 Mt) were the world’s leading producers of mined copper in 2018. China led the world in copper smelter and refinery production followed by Chile and Japan. For 2017, the Government of China’s most recently published production data were 1.7 Mt for mine, 6.4 Mt for smelter, and 8.9 Mt for refinery (China Nonferrous Metals Industry Association, 2019). For this study, China’s production data collected by facility totaled 886,000 metric tons for mine, 5.2 Mt for smelter, and 6.7 Mt for refinery; thus, of China’s reported 2017 production, mine was 53 percent of the total, smelter was 81 percent of the total, and refinery was 75 percent of the total. Because of the numerous small unknown facilities, it was difficult to locate and collect production data for many facilities in China. A large portion of the unrepresented facilities were mining operations. The unrepresented facilities are evident in table 1 where, using data collected for this study, China is ranked fifth in global copper mine production but is estimated by the USGS as the world’s third largest producer (Flanagan, 2021). As facilities were excluded from the dataset because of a lack of available production data, location data, or both, the risk estimates generated in this study likely underestimate the total risk.

Global Rhenium Capacity and its Geospatial Concentration

Annual global rhenium production by facility is not publicly available because of the small number of producers (n=12) and the extremely high value of rhenium; the 2018 annual average price of rhenium metal pellets (minimum 99.9 percent rhenium) was $1,470 per kilogram of contained rhenium (Polyak, 2023). Rhenium recovery facilities and their capacities, which include smelters and roasters, were estimated using USGS and industry publications (QY Research Inc., 2019; Roskill Information Services Ltd., 2015, 2019; Xun, 2020); where discrepancies were found, USGS compilations were used that were based on direct communications with relevant stakeholders. These estimated capacities were used as a proxy for rhenium production. The locations were verified and geolocated using Google Earth (Google LLC., 2019) and stored in a geodatabase. Total capacity estimated for the 12 rhenium facilities identified in this study was about 72 metric tons (fig. 2).

Global Earthquake Hazard and its Spatial Concentration

The Global Earthquake Model (GEM) Foundation has recently developed a Global Seismic Hazard Map (fig. 3) from a mosaic of regional (for example, South and Central America, Central Asia, sub-Saharan Africa, and Europe) and national (for example, Canada, the United States, Indonesia, Australia, New Zealand, Philippines, and South Africa) earthquake hazard models (Pagani and others, 2020). Most of the regional models were developed by GEM through collaboration with local earthquake scientists and engineers. Most of the national models were produced by agencies that are officially mandated by their respective governments to develop or update seismic hazard information for their country and were directly incorporated into the global mosaic. A minor fraction of the GEM Global Seismic Hazard Mosaic (Pagani and others, 2018) contains models that are available through the peer-reviewed literature (for example, India from Nath and Thingbaijam, 2012) or are developed internally at the GEM secretariat (for example, Russia and the Pacific Islands). The GEM Global Seismic Hazard Mosaic represents the state of the art based on openly available probabilistic seismic hazard assessment models. The hazard model used in the present study for South America is based on GEM’s global mosaic, whereas the Schnebele and others (2019) study relied on the South America hazard model from Petersen and others (2018), which was the latest publicly available hazard model at the time.

From the GEM Global Seismic Hazard Mosaic, we use a seismic hazard curve at each copper- or rhenium-producing facility that forecasts annual frequencies of exceeding each in a range of earthquake-induced peak ground accelerations (for example, refer to fig. 1 in Schnebele and others, 2019). From such hazard curves, the peak ground accelerations at an annual exceedance frequency of 1 in 475 years (in other words, a return period of 475 years) is mapped in figure 3. The map and hazard curves used here are for a reference rock condition because such curves are not yet available for other soil conditions from the GEM Global Seismic Hazard Mosaic. The map illustrates that the earthquake hazard is relatively high in Chile, Peru, Japan, and parts of China compared to other locations in this study.

Analysis of Global Supply Disruption Risk

As detailed in Schnebele and others (2019), the risk of earthquake disruption to mineral commodity supplies can be quantified through an application of the total probability theorem that combines the production at different facilities and a corresponding site-specific seismic hazard curve. For each facility, the method calculates an expected amount of mineral production disrupted by earthquakes per year, denoted as “EAD.” This calculation also integrates a model of the expected vulnerability of the facility production to earthquake shaking (in other words, the expected percentage of annual mineral production disrupted as a function of the shaking intensity at the facility). The results use two illustrative vulnerability models developed in Schnebele and others (2019), and each model is applied to all the facilities. The disruption results from the two vulnerability models represent two possible quantifications of the seismic risk to mineral production. The two models are loosely consistent with gathered reports on damage and disruptions to copper facilities caused during nine large South American earthquakes (Schnebele and others, 2019). The method of calculating EAD allows for facility-specific vulnerability models from future research. Regardless, the EAD results can be mapped to illustrate the spatial concentration of risk and summed across facilities (for example, by country).

Expected Annual Disruption of Copper

The analysis indicates that the EAD is about 0.3–1.1 percent of global copper mine production, about 1.8–4.0 percent of smelter production, and about 1.5–3.3 percent of refinery production, depending on the vulnerability model selected (table 2). Aggregated on a country level, the analysis indicates that Chile has the highest seismic risk for mine production (table 2; fig. 4A), which is expected for the largest producer of mined copper (table 1), mainly originating from a few facilities that are in a high seismic hazard region (fig. 3). Even though China led the world in copper smelter and refinery production, Japan is in a high seismic hazard region and has the greatest risk in terms of smelter and refinery production (table 2; fig. 4B, C). The two illustrative vulnerability models provided the same pattern in terms of relative risk among different countries. However, the vulnerability model A risk estimates, given higher susceptibility to production disruption for a given shaking intensity when compared with vulnerability model B, are roughly 2–5 times higher than the estimates obtained from vulnerability model B.

At the level of individual facilities, the results indicate that nine mines in Chile, Peru, Indonesia, and Papua New Guinea (fig. 4A) and only three (or four depending on the vulnerability model used) smelters and refineries in Japan (fig. 4B, C) account for about half of the total EADs. For copper mines, the operations that had the highest EADs have large production quantities and are in areas of high seismic hazard (fig. 5). Copper mines that had the greatest EADs are also operations that had predominately porphyry copper deposits. Porphyry copper deposits form along convergent plate boundaries and are in areas that are susceptible to earthquake hazards. Overall, these handful of operations, especially for the Japanese smelters and refineries, are of most concern from the perspective of copper supply disruption from earthquakes.

Expected Annual Disruption of Rhenium

The analysis indicates that the EAD is 0.32–1.32 percent for rhenium production capacity, depending on the vulnerability model selected (table 3). With the largest rhenium production capacity (fig. 2) and high seismic hazard (fig. 3), Chile faces the highest seismic risk in terms of total production disruption. Despite larger production capacity elsewhere (including Poland and the United States), the rest of the global (total) EAD is mostly in South Korea, where the seismic hazard at the facilities is higher. Also because of the spatial concentrations of seismic hazard, the relatively small production capacity in Armenia has risk comparable to that in the United States.

Effects of Annual Expected Disruption

Although market prices of copper and rhenium may not necessarily reflect the unit prices that producers would realize from their sales, they can provide a general proxy. At 2018 prices of copper (about $6.5 per kilogram from annual average London Metal Exchange, grade A, cash price; Flanagan, 2021) and rhenium ($1,470 per kilogram of 99.9 percent pure metal pellets; Polyak, 2023), lost revenue from the EAD totals ranges from $315 to $1,290 million for copper mining, $1.92 to $4.33 billion for copper smelting, $2.06 to $4.52 billion for copper refining, and $337,000 to $1.40 million for rhenium. All other things being equal, these effects will be greater for countries that have greater EAD. As illustrated in figure 6A, B, and C, countries that have higher copper production generally have greater EAD. Because of the differences in the earthquake hazard, however, some countries will have notably greater EAD at the same level of copper production. For example, Sweden and Papua New Guinea have roughly the same level of copper mine production. The copper mine EAD (using vulnerability model A) for Papua New Guinea is, however, more than 1,000-fold greater than that of Sweden (fig. 6A). Similar disparities exist among other countries and for the different copper production stages. Additionally, because of differences in each country’s economic profile, copper disruptions to each country’s economy will also differ. For some countries, including Chile, the Democratic Republic of the Congo, Mongolia, and Zambia, copper production provides a substantial contribution to their gross domestic product (GDP; fig. 6A, B, and C); therefore, copper production disruption is an important concern for these countries compared with other countries that have more diverse economies. The ability for facilities that are not affected by an earthquake to increase their production to offset these production losses will decrease these economic effects, at least for the global market.

The extent of damages may not necessarily stop at the producer; downstream manufacturers and final consumers may potentially have to pay higher prices for these commodities, reduce their output, or use less-preferred substitutes. The effects may be most acute for countries that are highly import reliant on copper and rhenium from countries that have the greatest EAD. Countries have varying degrees of import reliance. Moreover, they may source their raw materials from different locations and in different forms. As a result, the effects of supply disruptions may vary notably for many importing countries. For example, as a major copper producer, the United States is not highly net-import reliant on copper. Most of the unmanufactured copper that the United States imports is in the form of refined copper from Chile. In contrast, Russia imports most of its unmanufactured copper from Kazakhstan in the form of copper concentrates, which feeds its smelters and refineries. Thus, the potential effects of copper supply disruptions on the manufacturing sectors of the United States and Russia are expected to differ. However, because copper is a globally traded commodity, a supply disruption in one country may affect consumers either directly or indirectly because of higher prices. Additionally, inventories held by producers, consumers, merchants, and exchanges can help offset the effects of the disruptions. For copper, known inventories were greater than 1.1 Mt as of May 2023 (Kotseras, 2023), indicating that they would be more than sufficient in covering the EADs estimated based on the two candidate vulnerabilities (models A and B). If, however, the disruption levels exceeded the available inventories, an economic model would be necessary to determine the full extent of the economic damages that these disruptions would have to the specific industries and national and global economies overall.