NSERC/COSIA Industrial Research Chair in

Oil Sands Tailings Geotechnique

Development of an integrated tailings simulation model using System

Dynamics and GoldSim

Tony Zheng

• Introduction: System Dynamics and Causal

Loop Diagrams (CLD)

• Case Study: Soil Water Dynamics of Tailings Cap

- Part I: Modelling Process

- Part II: Results

• Concluding Remarks

2

Agenda

The Modelling Universe

3

Finite Element

Finite Difference

GEOTECHNICAL ENGINEERING

SOCIAL-ECONOMIC , ENVIRONMENT AND BUSINESS

Discrete Element

System Dynamics- Global Rules- Global Dynamics

Discrete Event- Global Rules- Local Dynamics

Agent Based- Local Rules- Local Dynamics

Hybrid Models

System Dynamics + Finite Difference

Stochastic

Non-Linearity

MY THESIS(Bobashev, 2015)

– Developed by Jay Forrester at MIT’s Sloan

Business School in the 50s:

To model complex inter-relationships between

elements within a system or multiple systems.

- Feedbacks, Feedbacks and Feedbacks

– Applications in Public Health, Management

Consulting, Water Resource Management,

Public Policy, International Relations, Defense

and Securities etc.

What is System Dynamics?

4

5

System Dynamics Modelling Process

Mental Model (s)

System Dynamics and Causal Loop Diagrams (CLD)

Simple Model

Detailed Model

Amount of Time Spent in the Real World

DiagramsQualitative

Quantitative

Transition

A Simple Example of Causal Loop Diagrams

6Modified from Sternman, 2000

R R: Reinforcing R

BB: Balancing

B

A Simple Example of Causal Loop Diagrams

7Modified from Sternman, 2000

Population

BR

R

B

B

R

A Simple Example of Causal Loop Diagrams

8 Modified from Sternman, 2000

Population ?

Soil Water Dynamics of Tailings Cap

Part I: Model Building Process

Part II: Simulation and Analysis

Case Study

10

Step 0 – Conceptualization

Coarse Sand Tailings Cap

Thickened Tailings

Evaporation

Run-Off

Precipitation

Upward Consolidation

Flux

3 m

50 m

11

Step 1: Pick the Governing Equations

Van Ganutchen - Maulem (1980) ; Wilson et al (1997); Huang et al (2011)

Vol Water Content

Hydraulic Conductivity

Relative Saturation

Inter-Layer Transmission Rate

Evaporation-Suction Model

Actual Evaporation Rate

Step 2: List relevant variables; identify the stock(s) and flows

Step 3: Construct the preliminary CLD;Identify feedbacks; discuss and debate

Step 4: Use the Bull’s Eye Diagram to identify the boundary of the model / system

EXOGENOUS

ENDOGENOUS

AE/PE Ratio

Suction Head

Vol Water Content

Unsat K

Water

Storage

PrecipitationPotential

Evaporation

Temperature

Void Ratio

Saturated K

Infiltration Rate

Rel Humidity

of Soil

Run-off

Thickness

Rel Humidity of Air

Consolidation Flux

EXCLUDEDSnowmeltVapour Transport

DiffusionFreeze/Thaw

OsmosisLateral

Flow

15

Step 5: Build the partial CLD; add polarity signs and identify feedback structures

15

Negative (Balancing) Loop

- Suction-Driven

- Dominates when

evaporation >

precipitation

Positive (Reinforcing) Loop

- Infiltration-Saturation Driven

- Dominates when

precipitation > evaporation

or when wetting front arrives

16

Step 6: Transform CLDs into GoldSim

16

17

Step 7: Validation

0

50

100

150

200

250

300

350

400

450

0 50 100 150 200 250 300 350 400So

il W

ater

Sto

rage

(m

m)

Time (days)

0

10

20

30

40

50

60

70

80

90

0 10 20 30 40 50 60 70

Ch

ange

in S

oil

Wat

er S

tora

ge (

mm

)

Time (Days)

Measured

GoldSim

Step 7: Run and validate simple models under a variety of conditions and different materials.

Dynamic Boundary Condition

Extreme Conditions

Step 8: Visualization and User Interface

User Input Model Overview

Layer Setup

Real-Time Simulation Results

User sees what the

modeler sees

Step 9: Go back to the Bull’s Eye Diagram if required as part of the iterative and participatory modelling process

Final Steps

Step 10: Parameter Estimation (Case Study)

Step 11: Simulation and Sensitivity Analysis (Case Study)

20

Case Study

Soil Water Dynamics of Tailings Cap

Part I: Model Building Process

Part II: Simulation and Analysis

Global Dynamics– Total Water Storage

21

Vary Saturated Hydraulic Conductivity of CST

Initial SC of TT = 60%

10-Year

22

Global Dynamics– Total Water Storage

Vary Initial SC of TTSame CST Ks in all scenarios

23

Global Dynamics– Cumulative Runoff

What is the minimum initial SC of TT required to prevent consolidation release water from daylighting at the surface?

24

Local Dynamics - Layer 4 (Depth: 52 cm)

Dashed Line: Higher KsSolid Line: Lower Ks

Arrival of Wetting Front

Un-Saturated Quasi-Saturated

• Why System Dynamics?

– Feedback Structures

– Rigorous Qualitative Process

– Foster thinking and shared understanding

through participatory modelling exercises

– Ability to model soft variables

Concluding Remarks

25

• Limitations

– Poor Capture of Spatial Variation

– Over-Simplification

– Over-Complexity

– Complacency?

Concluding Remarks

26

27

Acknowledgements

- Dr Nicholas Beier

Software Support:

FSCA (Dr Silawat Jeeravipoolvarn)