Approach for Estimating a Probable Range of Pit Lake

Approach for Estimating a Probable Range of Pit Lake Concentrations for Mine Pits

with Sulfide Wall Rock

Sarah Doyle; Cory Conrad; Colleen Kelley; Steve Lange; Rick Frechette; Houston Kempton

August 14, 2014

Outline

• What is a probabilistic model? • Benefits using a probabilistic approach • Methods • Example presentation of results • Limitations and ongoing challenges • Summary

2

Session 2: Tailings Technologies

• Scaling of lab data (temp, grain size, accelerated weathering) • Thermodynamic database: assume constants are correct and that

our system is in equilibrium • Assume complete mixing of ponds • Complex mechanisms (cyanide attenuation, sorption) are simplified.

Some are not modeled (coprecipitation)

3

Limitations of Modeling

• We can model uncertainty using Monte Carlo methods (represent inputs as range of values to get range of outcomes)

• We can avoid scaling issues by using field testing results as much as possible

“All models are wrong, but some are useful” - George E.P. Box, statistician

What is a probabilistic model?

4

0.00

0.10

0.20

0.30

1 2 3 4 5 6 7 8 9 10 11 12Pr

obab

ility

Den

sity

Total Iron [mg/L]

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0 1 2 3 4 5

Pro

babi

lity

Initial Percent Sulfide [%]

0.00

0.05

0.10

0.15

70 71 72 73 74 75 76 77 78 79 80

Pro

babi

lity

Pan Evap Coefficient [%]

Pro

babi

lity

INPUTS OUTPUTS

Benefits of Probabilistic Models

• Probabilistic models may reduce comments on EISs by allowing parties to “agree to disagree”

• Why is this important? – Rosemont (AZ): >43,000 comments; – NorthMet (MN): >50,000 comments; – Average EIS contractor cost for the DOE in 2013:

$2.9 million* – Federal Agencies are moving towards requiring EIS’s

for changes to existing projects

5 *U.S. Government Accountability Office, National Environmental Policy Act: Little Information Exists on NEPA Analyses, April 2014


• Increases understanding of project risks – Helps with: closure

cost estimates, alternatives analyses, etc.

• Worst-case scenario isn’t always obvious for complex, dynamic systems

6


7 Data from INAP Pit Lake Database

Modeling Methods

8

MODELING

Conceptual Water Balance

9

Evaporation

PIT LAKE

Direct Precipitation

Pit Wall Runoff

Groundwater

BEDROCK

Other Inflows

Conceptual Solute Balance

10

Damaged Rock Zone

Other Inflows

Pit Wall Runoff

Groundwater Oxidation of Sulfide Wall Rock

Evapo-Concentration

Pit Wall Runoff

11

𝑃𝑃𝑃 𝑊𝑊𝑊𝑊 𝑅𝑅𝑅𝑅𝑅𝑅 = 𝐶𝑅𝑅𝐶𝐶𝑅𝑃𝐶𝑊𝑃𝑃𝑅𝑅 ∗ 𝐹𝑊𝑅𝐹 𝑅𝑊𝑃𝐶 + 𝑆𝑅𝑊𝑅𝑃𝐶𝑆 𝑅𝐶𝑊𝐶𝑊𝑆𝐶𝑅 𝑅𝐶𝑅𝑓 𝑆𝑅𝑊𝑅𝑃𝑅𝐶 𝑂𝑂𝑃𝑅𝑊𝑃𝑃𝑅𝑅

Session 2: Tailings Technologies

• Data available for modeling depends on the phase of the project.

12

Available Data

Static Tests • ABA • Whole rock • Leach Tests

(MWMP, SPLP)

• NAG

Kinetic Tests • Humidity

cells

Field Tests • Field barrels • Test plots

Field Data • Site samples

Better Modeling Data

Kinetic Program Comparing FB and HC

• Field barrel acidity greater and faster • Humidity cells have such high flush rates that they do not develop “hot spots”

or acidic micro-environments where the sulphide oxidizing bacteria thrive.

2

3

4

5

6

7

8

9

10-Aug-10 7-Jan-11 6-Jun-11 3-Nov-11 1-Apr-12

Lab

pH

COL 28 - EBX2 (S=1.86%)

HC 1 EBX2 (S=2.04%)

COL 48 - IBX(S=2.10%)HC 2 IBX (S=2.16%)

COL 55-01 - DA(S=1.26%)HC 3 DA (S=1.26%)

COL 19 - H (S=1.12%)


• Greater flush of alkalinity in the humidity cells than in the field barrel due to high water infiltration rates and volumes

0.001

0.01

0.1

1

10

100


Alka

linity

(mg/

kg/w

k) COL 28 - EBX2

(S=1.86%)HC 1 EBX2 (S=2.04%)

COL 48 - IBX(S=2.10%)HC 2 IBX (S=2.16%)

COL 55-01 - DA(S=1.26%)HC 3 DA (S=1.26%)


• Greater sulphate release in the humidity cell tests than in field barrels due to lack of secondary mineral precipitation

0.1

1

10

100

1000


SO4

(mg/

kg/w

k)

COL 28 - EBX2 (S=1.86%)

HC 1 EBX2 (S=2.04%)

COL 48 - IBX (S=2.10%)

HC 2 IBX (S=2.16%)

COL 55-01 - DA (S=1.26%)

HC 3 DA (S=1.26%)

COL 19 - H (S=1.12%)

HC 4 H (S=1.12%)

Solute Release from Sulfide Oxidation

16

• Rate of sulfate formed from oxidation of sulfide wall rock was estimated based on humidity cell data.

• Defined in model by

slope and y-intercept


17

• Solute assumed to be released from sulfide oxidation if statistically significant correlation exists between sulfate and the solute in NAG Leachate

• Solute release modeled as ratio to sulfate release.


18

• If solute is not correlated to sulfate concentrations in NAG leachate, then it is not released with sulfide oxidation in the model.

• Solute can still be released from pit wall runoff based on leach test results.

Damaged Rock Zone Properties

19

DAMAGED ROCK

Parameter Min Max

Thickness1 3 m 15 m

Size Factor2 0.05 0.2

Percent in Contact with Runoff3

30% 90%

Sulfate Leach Rate From Humidity Cell Data

1Radian 1997; McClosky et al. 2003 2Malmstrom et al. 2000; Lopez et al. 1997 3Frostad et al. 2005; Hollings et al. 2001

Model Input

20

• Inflow concentrations are defined by probability distribution, in this case normal distribution, defined by mean and standard error of the mean

3.1416

MW_Stderr

3.1416

MW_Mean MW_Conc

Solute Balance Simulations

21

• GoldSim uses Monte Carlo method for propagating uncertainty in model inputs into uncertainties in model outputs.

• Simulations conducted several thousand times (realizations)

• For each realization, a different parameter value is selected from the input probability distribution.

0

1

2

3

4

5

6

0 10 20 30 40 50 60 70 80 90 100(m

g/L)

Time (yr)

Fluoride

Percentiles Greatest Result 95% Percentile 75% Percentile 25% Percentile 5% Percentile Least Result Mean Median

Geochemical Controls

22

• Bulk concentrations from the GoldSIM are analyzed for geochemical controls using PHREEQC:

• Atmospheric O2 and CO2 • Solubility controls for

probable mineral phases in pit lakes*

• Sorption to precipitated ferrihydrite

*Eary, L.E., Geochemical and equilibrium trends in mine pit lakes, Applied Geochemistry 14 (1999) 963-987

Summary Chemical Processes

Model Results

23

Model Results

24

Model Results

25

Model Results

26

Model Results

27

Summary “ Doubt is not a pleasant condition, but certainty is absurd”

- Voltaire

• Focus should be on capturing the uncertainty • Oxidation of sulfide wall rock is a large source

of uncertainty and commonly represented inaccurately

• Allows parties to “agree to disagree”

28

Ongoing Challenges

29

• When you set a percentile, there won’t be any interest in lower percentiles.

• 90th Percentile precedent:

• Idaho Cobalt EIS • PolyMet EIS • Ecological Risk Assessment Guidance

References (1 of 2)

30

Eary, L.E., 1999. Geochemical and equilibrium trends in mine pit lakes, Applied Geochemistry 14, pp. 963-987 Frostad, S., Klein, B., and Lawrence, R.W. 2005. Determining the weathering characteristics of a waste dump with field tests. Int. J. of Surface Mining, Reclamation and Environment, 19(2), pp. 132 – 143. Hollings, P., Hendry, M.J., Nicholson, R.V., and Kirkland, R.A. 2001. Quantification of oxygen consumption and sulphate release rates for waste rock piles using kinetic cells: Cluff Lake uranium mine, northern Saskatchewan, Canada. Applied Geochemistry, Vol. 16, pp. 1215-1230. International Network for Acid Prevention, Pit Lake Database <http://pitlakesdatabase.org/database/home.asp> Lopez, D.L., Smith, L. and Beckie, R. 1997. Modeling water flow in waste rock piles using kinematic wave theory. In Proceedings of the Fourth InternationalConference on Acid Rock Drainange, Vancouver, B.C. Canada, May 31 – June 6, 1997. Vol. 2, pp. 497 – 513.

References (2 of 2)

31

Malmstrom, M., Destouni, G., Banwart, S., and Stromberg, B.H.E. 2000. Resolving the scale-dependence of mineral weathering rates. Env. Sci. and Technol, 34(7), pp. 1375 – 1378. McClosky, L., R. Wilmoth, J.LeFever, and D. Jordan, 2003. Evaluation of technologies to prevent acid mine drainage generation from open pit highwalls, In: Proceedings of the 6th International Conference on Acid Rock Drainage, Cairns, Australia, 14-17 July, 2003, pp. 541-547. Radian, 1997. Predicted water quality in the Betze-Screamer pit lake. Prepared for Barrick Goldstrike Mines, Inc., Elko, NV. Prepared by Radian International. U.S. Government Accountability Office, 2014. National Environmental Policy Act: Little Information Exists on NEPA Analyses, April 2014.

Documents

Approach for Estimating a Probable Range of Pit Lake