and
@Thalamus, I've been refraining from a direct comparison between ARR and RDM since ARR's scoping study is due next month, which would make it much easier, but given MEI's scoping study has slipped to Q2, there's every chance I will be waiting 3-4 months. So without going too much into the details, the best way to estimate CAPEX and OPEX is to find studies (preferably at DFS and recent) that are at similar throughput and flowsheet.
The ARR Halleck Creek flowsheet has two components - the beneficiation plant and the rare earth leaching plant. The beneficiation plant use fine crushing, milling, DMS and WHIMS, so quite similar to spodumene processing and quite a few examples to draw on there. Their proposed leaching plant is very similar to a typical clay tank leach, so Aclara's Penco is probably the best analogue. The main difference is ARR is using 90degC and 250 kg/t acid leaching so the entire leach train and filtration circuit will need to be made from a lean duplex steel - about twice as much CAPEX for this equipment as painted carbon steel used for ambient pH 1-4 leaching.
I am going to assume ARR use a 5 mtpa ore throughput. At around double the head grade of RDM, this is equivalent to a 10 mtpa Sybella heap leach on feed magnet REO into the plant for both cases.
CAPEX for Beneficiation
I am going to draw on LTR's Kathleen Valley (KV) DFS for CAPEX here -
https://wcsecure.weblink.com.au/pdf/LTR/02450567.pdf. KV DFS is 2.5 mtpa, so I am going to double the CAPEX to estimate ARR at 5 mtpa (noting KV DFS is 2021, so inflation will eat away at any scale up efficiency). Below the comparison of the flowsheets to show how similar ARR's is with KV:
KV, with same unit ops as ARR circled. The only difference is KV has flotation, but ARR has HPGR rolls (higher capex than Jaw Crushers). Disclaimer - there will be different unit flows through each equipment.
KV CAPEX (not including mining which is contracted) was $473m AUD for 2.5 mtpa. They include a lot of infrastructure, so removing CAPEX as ARR may not require this, I am going to be generous to ARR and estimate $300m AUD CAPEX for 2.5 mtpa with their circuit and location. Scaling up to 5 mtpa and accounting for KV being 2021 costs, gets you to $600m AUD CAPEX for the beneficiation circuit for ARR.
CAPEX for Leaching
The feed into leaching for ARR is 7 weight units, or 7% x 5 mtpa = 0.35 mtpa. Scaling directly off of Aclara Penco's 1.7 mtpa plant using the 0.7 rule gets a CAPEX of ~$90m AUD for the leaching circuit assuming a 50% CAPEX increase from the use of 90degC leaching (probably still underestimated).
So total CAPEX for ARR in the vicinity for
$700m AUD for a 5 mtpa ore throughput plant. Compare with Sybella, where I have estimated $300m AUD for a 10 mtpa plant, both plants having the same feed throughput of rare earths due to differing head grades. So capital intensity for ARR is significantly higher per unit REO produced = IRR will be much lower for ARR vs RDM.
OPEX
OPEX is too difficult to estimate as ARR met data is very thin on the ground, but I will throw some numbers out. For the benefication circuit alone, KV had a total OPEX (fixed+variable) of $28.24 AUD/t ore feed for beneficiation. Assume 0 strip ratio hard rock mining $4.50/t for ARR. Already this comes to ~$33 AUD/t ore for mining and beneficiation (no leaching circuit), compared to RDM around $22/t ore. So ARR's OPEX may be on par with RDM per unit of REO produced.
Final note,
testwork DMS results never match full scale reality. ARR's testwork for DMS uses a heavy liquid at a set SG. During the test the material is allowed to reach equilibrium in the heavy liquid to get a perfect scenario recovery and grade cut at that SG. Reality with cyclones is never perfect, and Core (CXO)'s Finniss plant demonstrates this; DFS Li recovery using DMS circuit based on testwork = 71.7%. Full scale Li recovery using DMS after 6 months of commissioning = 50%. That's a massive drop going from laboratory testwork (perfect recovery-grade cut) to real life.
(20min delay)