Steps toward Multi-Asset Treatment Optimization
24
Fritz Jooste Lonrix Ltd Steps toward Multi-Asset Treatment Optimization DUNEDIN 2019 ROAD INFRASTRUCTURE MANAGEMENT FORUM
Transcript of Steps toward Multi-Asset Treatment Optimization
Slide 1D U N E D I N
2019 ROAD
• Co-Author: KarunaKar Kintada (KyuGIS and surveys)
• Co-Author: Sanet Jooste (Juno Services)
• Colin Scott (Waikato District Alliance)
• Phillipa O’Shea (Downer)
• Sean Rainsford (Fulton Hogan)
Steps toward Multi-Asset Treatment Optimization
ROAD INFRASTRUCTURE MANAGEMENT FORUM DUNEDIN
Different levels of approaching the challenge:
Steps toward Multi-Asset Treatment Optimization
ROAD INFRASTRUCTURE MANAGEMENT FORUM DUNEDIN
Cross Asset Treatment Optimization
Strategic Viewpoint Operational Viewpoint
Key Issue: Rational way of optimal budget
assignment amongst asset groups Key Issue: Rational way of optimal application
of resources, given existing works plans
Solving the challenge:
Strategic Cross Asset Optimization
1. Historical or Formula Applications (e.g. based on historical percentages)
2. Performance based (funding tied to gaps between current and target performances)
3. Optimization Methods
infrastructure management (PhD Thesis, University of Texas at Austin)
Key Steps in Strategic Cross Asset Optimization:
• Formulation of agency objectives and KPIs
• A common performance scale to compare performance across asset groups
• An agency-wide score or performance indicator is needed
• Formulation of solution methodology or approach
• Modelling and analysis
An Example:
Assumptions: • A calibrated deterioration
• Performance-vs-budget data
Assumptions: • A calibrated deterioration
• Performance-vs-budget data
Assumptions: • A calibrated deterioration
• Performance-vs-budget data
group
• Footpaths for 100% = 25
perfect results
An Example:
Problem Statement:
• Total needs for perfect asset condition in all groups is 110 • If we have a budget of only 80 (i.e. 72% of desired), how to divide this amongst the three asset groups?
Points to Note: • Roads needs for 100% = 50
• Water needs for 100% = 35
• Footpaths for 100% = 25
Three Approaches: • Maximize Utility
Available Total Budget = 80
Strategic Cross Asset Optimization
Need for 100% Performance:
Roads Water Footpaths Total
50 35 25 110
Utility ( = Budget/Need): 0.80 0.71 0.60 2.1
Envy (comparing Utility): 0.40.0
Roads to Water = 0
Roads to Footpaths = 0
Variables
Objective
Functions
Constraint
Strategic Cross Asset Optimization
Objective Function Roads Water Footpaths Perf-Raw Utility-Total Utility-Min Envy-Utility Envy-Perf
Allocated Budget Objective Function Values
36.4 25.5 18.2Minimize Envy-Utility
Minimize Envy-Perf
30.2 33.0 16.8 24.1 2.22 0.60 0.68 0.0
25.5 2.20 0.6 0.8 5.1
Maximize Total Utility 20.0 35.0 25.0 20.0 2.4 0.4 1.2 20.0
Maximize Minimum Utility 36.4 25.5 18.2 21.0 2.18 0.73 0.00 10.6
We can also optimize on:
Total performance of all three asset groups vary depending on the objective function
Remember:
Need for 100% Performance: Roads = 50; Water = 35; Footpaths = 25;
Perfect score for all 3 asset groups = 30
Elitist
Egalitarian
Observations:
• Objective function chosen greatly influences outcome (social vs engineering)
• Not a linear optimization problem (cannot use Simplex Method)
• Starting values can influence outcome (“local optimum”)
• Study of “best” objective function is needed
Software Solution (in place within 2019):
Strategic Cross Asset Optimization
Required Inputs
KPI Formulation (area under curve,
value at end)
Objective Function (performance/
Automatically run budget sensitivity over models for all asset types
Extract KPIs and automatically formulate Performance Curves
Run Optimization
Different levels of approaching the challenge:
Steps toward Multi-Asset Treatment Optimization
ROAD INFRASTRUCTURE MANAGEMENT FORUM DUNEDIN
Cross Asset Treatment Optimization
Strategic Viewpoint Operational Viewpoint
Key Issue: Rational way of optimal budget
assignment amongst asset groups Key Issue: Rational way of optimal application
of resources
at this component
better manage different works plans with
temporal and spatial alignments
Workflow of this Functionality:
Operational Cross Asset Management
• Choose a “Master Works Programme” (e.g. roadworks)
• Scan all works programs (drainage, water-cutting, line-marking etc)
• Assess time and geospatial proximity amongst programs
• Cumulatively score each work area based on “potential alignment score”
• Innovatively display areas with a high score (map or export)
Potential Alignment Score:
ROAD INFRASTRUCTURE MANAGEMENT FORUM DUNEDIN
User can define a custom score for time and space correspondence:
S c o
• User selects Master FWP and associated FWPs to consider
• Algorithm runs and shows results on a map or exports to Excel
“Hot Spot” indicator
potential to re-align works to share resources
Showing Outputs:
• User selects Master FWP and associated FWPs to consider
• Algorithm runs and shows results on a map or exports to Excel
Area Feedback:
Line-marking 2020
1. Strategic Cross Asset Optimization by means of Deterioration Modelling
2. Operational Cross-FWP Management by means of Geospatial-temporal methods
Summary and Conclusions
• Accurate models for individual assets are essential
• A common performance index for all assets, and a defined combined KPI are needed
• Workflow and optimization algorithms have been tested and validated using Excel
• Looking toward a release for industry by Nov 2019 ( depending on interest!)
Two approaches have been developed to respectively address:
1. Strategic Cross Asset Optimization by means of Deterioration Modelling
2. Operational Cross-FWP Management by means of Geospatial-temporal methods
For Operational Cross-FWP Management:
• Algorithm developed to effectively score treatment lengths for geospatial and temporal overlap
• User definable scores for time and proximity
• “Hot Spot” display on Network Map for areas with Resource Sharing potential
• Looking toward a release for industry by September 2019 ( depending on interest!)
Summary and Conclusions
Thank You!
In this presentation:
• Strategic optimization – historic approaches and concepts
• Strategic optimization – a worked example and outputs
• Strategic optimization – keys to success / implementation for NZ
• Operational optimization – approach and concepts
• Operational optimization – Scoring and display of outputs
• Conclusions
There is a strategic need for Multi-Asset treatment Optimization:
• This is an optimization problem,
• Cross Asset Optimization (CAO)
• Can apply at different strategic levels down to operational levels
Steps toward Multi-Asset Treatment Optimization
ROAD INFRASTRUCTURE MANAGEMENT FORUM DUNEDIN
There is an operational need for Multi-Asset treatment Optimization:
• Mainly a management problem, solved with business rules and operational systems
• Cross Asset Management (CAM)
Keys to this Solution:
Strategic Cross Asset Optimization
• Everything relies on accurate performance curves for each asset group!
• Formulation of performance using a common index is essential
• Proper formulation of objective function
2019 ROAD
• Co-Author: KarunaKar Kintada (KyuGIS and surveys)
• Co-Author: Sanet Jooste (Juno Services)
• Colin Scott (Waikato District Alliance)
• Phillipa O’Shea (Downer)
• Sean Rainsford (Fulton Hogan)
Steps toward Multi-Asset Treatment Optimization
ROAD INFRASTRUCTURE MANAGEMENT FORUM DUNEDIN
Different levels of approaching the challenge:
Steps toward Multi-Asset Treatment Optimization
ROAD INFRASTRUCTURE MANAGEMENT FORUM DUNEDIN
Cross Asset Treatment Optimization
Strategic Viewpoint Operational Viewpoint
Key Issue: Rational way of optimal budget
assignment amongst asset groups Key Issue: Rational way of optimal application
of resources, given existing works plans
Solving the challenge:
Strategic Cross Asset Optimization
1. Historical or Formula Applications (e.g. based on historical percentages)
2. Performance based (funding tied to gaps between current and target performances)
3. Optimization Methods
infrastructure management (PhD Thesis, University of Texas at Austin)
Key Steps in Strategic Cross Asset Optimization:
• Formulation of agency objectives and KPIs
• A common performance scale to compare performance across asset groups
• An agency-wide score or performance indicator is needed
• Formulation of solution methodology or approach
• Modelling and analysis
An Example:
Assumptions: • A calibrated deterioration
• Performance-vs-budget data
Assumptions: • A calibrated deterioration
• Performance-vs-budget data
Assumptions: • A calibrated deterioration
• Performance-vs-budget data
group
• Footpaths for 100% = 25
perfect results
An Example:
Problem Statement:
• Total needs for perfect asset condition in all groups is 110 • If we have a budget of only 80 (i.e. 72% of desired), how to divide this amongst the three asset groups?
Points to Note: • Roads needs for 100% = 50
• Water needs for 100% = 35
• Footpaths for 100% = 25
Three Approaches: • Maximize Utility
Available Total Budget = 80
Strategic Cross Asset Optimization
Need for 100% Performance:
Roads Water Footpaths Total
50 35 25 110
Utility ( = Budget/Need): 0.80 0.71 0.60 2.1
Envy (comparing Utility): 0.40.0
Roads to Water = 0
Roads to Footpaths = 0
Variables
Objective
Functions
Constraint
Strategic Cross Asset Optimization
Objective Function Roads Water Footpaths Perf-Raw Utility-Total Utility-Min Envy-Utility Envy-Perf
Allocated Budget Objective Function Values
36.4 25.5 18.2Minimize Envy-Utility
Minimize Envy-Perf
30.2 33.0 16.8 24.1 2.22 0.60 0.68 0.0
25.5 2.20 0.6 0.8 5.1
Maximize Total Utility 20.0 35.0 25.0 20.0 2.4 0.4 1.2 20.0
Maximize Minimum Utility 36.4 25.5 18.2 21.0 2.18 0.73 0.00 10.6
We can also optimize on:
Total performance of all three asset groups vary depending on the objective function
Remember:
Need for 100% Performance: Roads = 50; Water = 35; Footpaths = 25;
Perfect score for all 3 asset groups = 30
Elitist
Egalitarian
Observations:
• Objective function chosen greatly influences outcome (social vs engineering)
• Not a linear optimization problem (cannot use Simplex Method)
• Starting values can influence outcome (“local optimum”)
• Study of “best” objective function is needed
Software Solution (in place within 2019):
Strategic Cross Asset Optimization
Required Inputs
KPI Formulation (area under curve,
value at end)
Objective Function (performance/
Automatically run budget sensitivity over models for all asset types
Extract KPIs and automatically formulate Performance Curves
Run Optimization
Different levels of approaching the challenge:
Steps toward Multi-Asset Treatment Optimization
ROAD INFRASTRUCTURE MANAGEMENT FORUM DUNEDIN
Cross Asset Treatment Optimization
Strategic Viewpoint Operational Viewpoint
Key Issue: Rational way of optimal budget
assignment amongst asset groups Key Issue: Rational way of optimal application
of resources
at this component
better manage different works plans with
temporal and spatial alignments
Workflow of this Functionality:
Operational Cross Asset Management
• Choose a “Master Works Programme” (e.g. roadworks)
• Scan all works programs (drainage, water-cutting, line-marking etc)
• Assess time and geospatial proximity amongst programs
• Cumulatively score each work area based on “potential alignment score”
• Innovatively display areas with a high score (map or export)
Potential Alignment Score:
ROAD INFRASTRUCTURE MANAGEMENT FORUM DUNEDIN
User can define a custom score for time and space correspondence:
S c o
• User selects Master FWP and associated FWPs to consider
• Algorithm runs and shows results on a map or exports to Excel
“Hot Spot” indicator
potential to re-align works to share resources
Showing Outputs:
• User selects Master FWP and associated FWPs to consider
• Algorithm runs and shows results on a map or exports to Excel
Area Feedback:
Line-marking 2020
1. Strategic Cross Asset Optimization by means of Deterioration Modelling
2. Operational Cross-FWP Management by means of Geospatial-temporal methods
Summary and Conclusions
• Accurate models for individual assets are essential
• A common performance index for all assets, and a defined combined KPI are needed
• Workflow and optimization algorithms have been tested and validated using Excel
• Looking toward a release for industry by Nov 2019 ( depending on interest!)
Two approaches have been developed to respectively address:
1. Strategic Cross Asset Optimization by means of Deterioration Modelling
2. Operational Cross-FWP Management by means of Geospatial-temporal methods
For Operational Cross-FWP Management:
• Algorithm developed to effectively score treatment lengths for geospatial and temporal overlap
• User definable scores for time and proximity
• “Hot Spot” display on Network Map for areas with Resource Sharing potential
• Looking toward a release for industry by September 2019 ( depending on interest!)
Summary and Conclusions
Thank You!
In this presentation:
• Strategic optimization – historic approaches and concepts
• Strategic optimization – a worked example and outputs
• Strategic optimization – keys to success / implementation for NZ
• Operational optimization – approach and concepts
• Operational optimization – Scoring and display of outputs
• Conclusions
There is a strategic need for Multi-Asset treatment Optimization:
• This is an optimization problem,
• Cross Asset Optimization (CAO)
• Can apply at different strategic levels down to operational levels
Steps toward Multi-Asset Treatment Optimization
ROAD INFRASTRUCTURE MANAGEMENT FORUM DUNEDIN
There is an operational need for Multi-Asset treatment Optimization:
• Mainly a management problem, solved with business rules and operational systems
• Cross Asset Management (CAM)
Keys to this Solution:
Strategic Cross Asset Optimization
• Everything relies on accurate performance curves for each asset group!
• Formulation of performance using a common index is essential
• Proper formulation of objective function
