FlagstoneReport
Does the work support the claim?
Assessment of an NI 43-101 technical report

Cascabel Project (Alpala and Tandayama-America porphyry Cu-Au deposits)

Imbabura Province, northern Ecuador (~100 km north of Quito)
Study stage
PFS · Pre-Feasibility Study
Published
March 2024 (Mineral Reserve effective date 31 December 2023; Mineral Resource effective date 11 November 2023)
Prepared by
SRK Consulting (Canada) Inc. for SolGold plc (QPs incl. Dr. Gilles Arseneau P.Geo., Jarek Jakubec C.Eng.)
Source
FS-PFS_Cascabel.pdf
Study-level anchor — not scored
PFS Pre-Feasibility Study

Every judgment below is measured against what a PFS is expected to demonstrate — not an absolute bar.

Detected from the filing

Report titled 'NI 43-101 Technical Report on Pre-Feasibility Study, Cascabel Project'; Section 1.1 states the purpose is to document work supporting a pre-feasibility study (PFS); Section 1.19 states the capital cost estimate 'meets the requirements for a PFS, consistent with AACE International cost estimating guidelines for a Class 4 estimate' with +/-25% accuracy. Mineral Reserves are declared, consistent with PFS-level confidence.

Peer Benchmarks

2 of the report's figures are benchmarked against PFS peers — shown as the expected range for the stage and where this project sits within it.

2 within range
Cu Recovery – Flotation88.5%

Sits within the expected range for PFS.

82%93.1%
Typical PFS range 86.3%–90.9% · full peer span 82%–93.1%
Capex Intensity – Underground / Block Cave$130 /t·yr

Sits within the expected range for PFS.

$43$170
Typical PFS range $106–$166 · full peer span $43–$170
Internal consistency
Recovery claim vs. testwork88.5 → 94.8%

The claimed 88.5% carried into the economics is backed by the 94.8% testwork result.

Sitting outside the expected range is context to investigate, not a verdict — a project can legitimately sit apart for geological or metallurgical reasons. These positions inform the reader; on their own they do not decide the grade.

Flagstone Score

We grade two crucial dimensions, then determine a final score from them.

BMeets

Methods robustness

Does the disclosed work match what the study stage demands?

Judged against the Part 3 PFS column: bench-scale flotation is present and extensive (three multi-year programs across nine labs); variability composites are far above the PFS expectation of 'several, across ore types' - 53 variability samples selected on lithology, alteration, Cu and Au grade, spatial position and mine-plan timing, each with mineralogy, comminution and open-circuit flotation; locked-cycle testing (expected at PFS) is present on three grade-class composites with recycle streams closed; mineralogy is present and linked qualitatively to metallurgy (98.1% of Cu in chalcopyrite, 55% average liberation at 150 um supporting the grind selection; refractory Au in pyrite and pH 11 pyrite depression explaining the depressed gold recoveries; kaolinite flagged as a recovery risk); the flowsheet is deposit-specific and justified by measured data (grind-recovery scan selecting F80 150 um, collector screening selecting PAX, frother change to polyglycol after poor froth quality, regrind target set by feed grade, cyanidation circuit tested and deliberately excluded). Pilot-scale work is not expected at PFS and its absence is not counted against the report. Two matters temper the letter: (i) the recovery model is a head-grade regression from LCT data with no metallurgical domaining despite mineralogical evidence (lithology-dependent competency, kaolinite, talc 'specific locations not quantified yet') that domains matter; and (ii) the QP states 'No corrections have been applied at this time for operability-related losses', so the design recovery is an unadjusted laboratory-derived value. On the Part 7 anchors the work satisfies every expected-for-stage item (B) and touches two A-axes (variability count well beyond the stage minimum; deposit-specific flowsheet justified by measured liberation and comminution data), but the undomained recovery framework and un-discounted LCT basis hold it at B rather than A.

  • Report omissionaffects the gradeThe recovery model is not segmented into metallurgical domains (lithology/alteration/mineralogy); grade classes (low/medium/high Cu) are used as a proxy, even though the report's own mineralogy identifies kaolinite and talc as potential recovery-impacting phases whose spatial distribution is 'not quantified yet'.
  • Report omissionaffects the gradeNo operability/scale-up discount is applied to the locked-cycle-derived recovery relationships ('No corrections have been applied at this time for operability-related losses'), so plant-basis recovery is not demonstrated at PFS level.
  • Data gapnot in the extracted text — does not affect the gradeMetallurgical testwork is disclosed for the Alpala underground deposit only; no flotation or comminution testwork is disclosed for the Tandayama-America deposit, which carries a separate Mineral Resource (including an open-pit-constrained portion). TAM is not in the current mine plan, so this may be addressed in sections not provided.
  • Data gapnot in the extracted text — does not affect the gradeThe plant design grind of P80 200 um (Summary 1.13) differs from the F80 150 um basis on which the flotation testwork and recovery equations were developed; the testwork showed recovery consistent from F80 150-200 um, but the design-basis reconciliation is not spelled out in the provided text (may appear in Section 17 Recovery Methods).
Metallurgical framework quality
Considers operating conditions beyond head gradepartial
Metallurgical domains defined, each with its own recoverypartial
Range of lab results explained by the frameworkpartial

Explained vs. unexplained The framework explains a substantial part of the observed lab range but not all of it. Copper feed grade plus target concentrate grade together account for the dominant trend: LCT recoveries move systematically from 80.4% (0.21% Cu feed) to 86.2% (0.57% Cu) to 94.8% (1.54% Cu), and the report explains the 32% of variability tests that failed the 22% Cu concentrate target by head grade (failures averaged 0.38% Cu feed versus 1.08% Cu for those that passed). Grind size, residence time, reagent regime, cleaner pH and regrind size were all optimised experimentally and the regrind target is set by feed grade, and Figure 13-4 makes recovery an explicit function of the targeted concentrate grade - so operating conditions are studied and partly embedded, but the production model itself is a recovery-versus-feed-grade curve rather than a condition-parameterised model. What remains unexplained is the within-grade-class scatter: at 0.4-0.9% Cu feed, 25 variability samples average 79.2% Cu recovery with 36% failing the concentrate target, and the overall open-circuit variability range is 63.4-94.4% Cu recovery and 30.8-87.4% Au recovery. The report identifies plausible causes (kaolinite, talc, lithology, Au refractory in pyrite) but does not build them into domains or quantify their spatial extent, so that residual spread is acknowledged rather than accounted for.

What the QP flagged in Section 13
  • limitationOperability losses not applied to recovery estimatesSection 13.7: 'The overall recovery relationships for the flotation plant are based on the available locked cycle testwork (LCT). No corrections have been applied at this time for operability-related losses.' The design and economic recoveries are unadjusted laboratory values; real plant recovery would normally be discounted, so the 88.5% LOM Cu recovery should be treated as an upper-bound basis at PFS.
  • riskConcentrate grade target not achieved by a third of variability samples32% of the 53 variability tests failed to achieve a copper concentrate above 22% Cu; failing samples averaged 0.38% Cu head versus 1.08% Cu for passing samples. Among medium-grade samples (0.4-0.9% Cu, 25 of 53 samples), 36% failed to reach 22% Cu concentrate, averaging 24.7% Cu concentrate at 79.2% Cu recovery. Meeting the 22% Cu concentrate target in lower-grade production years may require accepting lower recovery or lower concentrate grade, affecting both revenue and smelter terms in later mine life.
  • riskClay (kaolinite) contentMineralogy table observation: 'Where kaolinite is detected in quantity, recoveries may be impacted.' Other non-sulphide gangue ranges 1.1-19.0%. A named but unquantified recovery risk that is not carried into the recovery model or into domaining.
  • limitationTalc distribution unquantifiedTalc (0.1-0.3%) noted with the comment 'Specific locations not quantified yet.' Talc is a known flotation contaminant/naturally floating gangue; its spatial distribution remains an open item for later study stages.
  • riskRefractory gold in pyrite and pyrite depressionUTas microanalysis found gold hosted in refractory form, chemically bound in pyrite with compositional zoning and Au-rich inclusions; Au-Ag-tellurides may also be important. Cleaner flotation at pH 11 to suppress pyrite 'resulted in reduced gold recovery at finer grinds as gold associated with liberated pyrite struggled to float.' Gold recovery is structurally limited (LCT averages 48.5-75.5%; variability range 30.8-87.4%) and is a direct trade-off against copper concentrate grade.
  • limitationFroth quality in cleaner flotation'Poor froth quality was noted during cleaner flotation, which was overcome by replacing MIBC with a stronger polyglycol frother during locked cycle testing.' Identified and mitigated in testwork, but indicates cleaner-circuit sensitivity to reagent regime.
  • opportunityGold recovery from flotation tailings by cyanidationTen flotation-tails composites leached 46-88% (average 68.8%) of contained gold, and a refloated pyrite concentrate leached 76%, but cyanide consumption averaged 3.2 kg/t (12.19 kg/t for the pyrite concentrate) due to copper. The current preferred flowsheet has no gold cyanidation circuit. Incremental gold revenue was quantified and deliberately excluded on cost grounds; a potential upside if reagent consumption can be managed.
  • riskIncreasing ore competency with depth and decreasing gradeComminution results show 'as the head grade of copper decreases, the samples will generally be more competent and abrasive. The volcanic and D15 lithologies are typically more competent, and the deposit competency generally increases with depth.' Later mine life combines lower grade (lower recovery, harder concentrate-grade target) with harder, more abrasive feed - a compounding throughput/cost risk, mitigated in design by using 80th-percentile parameters.
BAdequate

Result representativity

Are the reported numbers trustworthy as reported?

Against the Part 4 signals the disclosure is unusually honest on variability: every key metallurgical parameter is reported as a range as well as an average (LCT recovery ranges per grade class, 53-sample variability ranges for Cu and Au recovery and concentrate grade, comminution ranges with 80th percentiles, concentrate deleterious-element minima/maxima). Poor-performing results are disclosed rather than suppressed - the report explicitly states that 32% of variability tests failed the 22% Cu concentrate target, and that 36% of medium-grade samples failed, and it reports the low-grade composite at 80.4% Cu and 48.5% Au recovery. Test conditions are fully disclosed (grind, pH, reagents, residence time, regrind size, recycle configuration) and the recovery basis is clearly stated as LCT-derived with the caveat that no operability correction has been applied. Recovery is grounded in mineralogy (Cu deportment to chalcopyrite, liberation at 150 um, refractory Au in pyrite explaining the gold shortfall). The headline 88.5% LOM Cu recovery is derived from and sits within the measured LCT envelope, and the year-1-10 / later-years split (90.8% / 86.8%) honestly shows recovery declining with head grade rather than presenting a single flat number. Reservations that hold it at B rather than A: (i) the Mineral Resource cut-off calculations for both Alpala and TAM use 93% Cu and 83% Au recovery, materially above the Section 13-derived 88.5% / 70%, with no reconciliation in the provided text; (ii) the Mineral Reserve note quotes 'metallurgical recoveries that range from 68-81% for copper and 85-92% for gold', which appears transposed and conflicts with Section 13; and (iii) the design grind of P80 200 um differs from the F80 150 um testwork basis. These are real internal inconsistencies but they sit in resource/reserve footnotes rather than in the metallurgical results themselves, and there is no evidence of cherry-picked conditions or hidden variability - so 'one or two real reservations' (B) rather than 'multiple representativity concerns' (C).

  • Report omissionaffects the gradeThe 93% Cu / 83% Au recoveries used in the Mineral Resource cut-off grade derivation are not reconciled against the Section 13 LOM recoveries of 88.5% Cu / 70% Au anywhere in the provided text.
  • Report omissionaffects the gradeThe Mineral Reserve footnote's stated recoveries (68-81% Cu, 85-92% Au) are inconsistent with, and appear transposed relative to, the Section 13 testwork results; no explanation is given.
  • Report omissionaffects the gradeNo sensitivity or downside case is presented showing the economic effect of the un-corrected (no operability loss) recovery basis.
  • Data gapnot in the extracted text — does not affect the gradeEconomic sensitivity to recovery, and the detailed production schedule recoveries year by year, would sit in Sections 21/22 which were not provided.
Flagstone Score
B

Work solidly supports the stage; minor reservations.

How the grade was reached

The letter is a deterministic roll-up of the two dimension grades and the peer flags — the same inputs always produce the same grade.

  1. Recovery: claimed 88.5% (carried into economics) vs measured 94.8% -> gap -6.3 pts
  2. Gap not material -> no representativity penalty
  3. Weakest link: methods=B, representativity=B -> take the lower = B
  4. Flag penalty: amber Opex/t milled ($/t) [opex] -0.25 (reduced: cost metric, provisional band)
  5. Weighted penalty = -0.25 notches, capped at -3 and floored to whole notches -> -0
  6. Final grade (floored at D) = B

Study level is unambiguously PFS (title page, AACE Class 4 +/-25% estimate, Mineral Reserves declared), so methods are judged against the PFS column of Part 3. METHODS - B (Meets). Every expected-for-stage item is present and several are well beyond the minimum: bench flotation across three multi-year programmes at nine independent laboratories; 53 variability samples characterised for mineralogy, comminution and open-circuit flotation, versus a PFS expectation of merely 'several, across ore types'; locked-cycle testing (the PFS requirement) completed on three grade-class composites with closed recycle streams; mineralogy present and used to explain metallurgy (chalcopyrite deportment and liberation supporting the 150 um grind; refractory gold in pyrite explaining gold losses); and a genuinely deposit-specific flowsheet built on measured data (grind-recovery scan, collector screening, frother substitution, grade-dependent regrind, tested-and-rejected cyanidation circuit). Pilot-scale work is not expected at PFS and is not counted as a gap. Two things keep this at B rather than A: the production recovery model is an undomained head-grade regression - grade classes stand in for metallurgical domains even though the report itself names kaolinite, talc (locations 'not quantified yet') and lithology as recovery-relevant - and the QP states no operability-loss correction has been applied to the LCT-derived recoveries. Under the Part 4B framework test, operating conditions are partially embedded (concentrate-grade trade-off is explicit; grind, pH, reagents and residence time were optimised but are not parameters of the model), domains are partial (grade bins, not mineralogical domains), and variability is partially explained (head grade explains the trend and the concentrate-grade failures, but the 63.4-94.4% open-circuit spread and the 36% medium-grade concentrate-target failure rate are named rather than resolved). At PFS that is acceptable-but-not-exceeding, so the framework holds the letter at B. REPRESENTATIVITY - B (Adequate). The metallurgical reporting is exemplary on the signals that matter: ranges behind every average, disclosure of failing tests, fully stated test conditions, an explicit statement of what the recovery basis is and is not, and mineralogical grounding for both the copper and gold numbers. The claimed LOM recovery of 88.5% Cu is derived from and sits inside the measured LCT envelope (80.4-94.8%), and the report shows recovery declining with grade over mine life rather than a flat figure. It is held at B by two unreconciled internal conflicts outside Section 13 - the 93% Cu / 83% Au recoveries used in both Mineral Resource cut-off derivations versus the 88.5% / 70% derived from testwork, and the Mineral Reserve footnote's apparently transposed 68-81% Cu / 85-92% Au recoveries - plus the design grind (P80 200 um) sitting above the testwork basis (F80 150 um). PEER POSITION (PFS/FS bands only): the 88.5% LOM flotation recovery sits below the 90.5% green threshold for Cu porphyry flotation recovery, i.e. it is not an outlier claim; opex of $15.24/t processed sits marginally above the $15.1/t green threshold; the design grind of P80 200 um sits at the boundary of the 150-200 um green range. Capex intensity is basis-sensitive for this phased project ($1,554M initial capital against 12 Mtpa Phase 1 versus 24 Mtpa fully expanded) and readers should note which basis any comparison uses. The single point a diligence reader should look at hardest is the recovery basis chain: an uncorrected laboratory recovery flowing into economics, an undomained grade-only recovery curve, and resource/reserve footnote recoveries that do not match Section 13.