Analytical Tools for Asset Management

1

USING ANALYTICAL TOOLS TO IMPROVE ASSET MANAGEMENT FOR T&D

May 5-7, 2004Hyatt Regency BostonBoston Massachusetts

Sponsored by EUCI

S. Chapel AssociatesLee Merkhofer Consulting

VMN Group LLC

2 2

Outline

• Introduction

• T&D Project Prioritization

• Managing Aging T&D Assets

• T&D Strategic Reliability

• Summary

OverviewPurpose of WorkshopUsing Models

ProblemDecisionsAnalytic ToolsDataOrganizational Issues

TheoryData & ModelsOrganizational Issues

1

Using Analytical Tools to Improve Asset Management for T&D

INTRODUCTION

Steve Chapel, S. Chapel Associates

Lee Merkhofer,Lee Merkhofer Consulting

Charles D. Feinstein and Peter A. MorrisVMN Group LLC

May 2004

2 2

Agenda

• Overview– Some Asset Management Beliefs: fact or fallacy?– History of our Asset Management Research – T&D Asset Management Issues

• Purpose of Seminar• Using Models

3 3

Some Asset Management Ideas – True or False?

• Problems can be solved by organizational change and asset management teams

• "Cost Benefit analysis, the "tried and true" method for ranking projects is a perfectly reasonable way to select projects to fund."

• The first important step is to gather data• All projects can be valued using the same aggregate

measures (i.e. $)• Projects are risky because of uncertain financial

consequences • Beta is an appropriate way to measure project risk

4 4

Some Asset Management Ideas – True or False?

• An important objective is near-term profitability • The biggest problem for asset management is

insufficient data. • ROI is a good metric for evaluating investments. • Ranking is a good way to prioritize projects. • Strategic alignment is a good metric for prioritizing

projects. • Balanced scorecards are a good way to prioritize

projects. • Hurdle rates are a good way to account for risks.

5 5

Agenda



6 6

T&D Research 1997 – 2002

• December 1996 Chicago Focus Group• 1997 - 1998

– Create the portfolio – distribution planning – Produce V1.0 Area Investment Planning Tools

• 1999– Start work on Customer Needs & Reliability projects– V1.5 Strategy Model & V2.0 LoadDynamics – Methodology design for Project Prioritization (AEP)

7 7

T&D Research 1997 – 2002 cont.

• 2000– Aging Assets - Started– Project Prioritization – method & software designed – Customer Needs & Reliability – EPRI white papers– Area Investment Planning – tech transfer

• 2001: Focus on Aging Assets and Project Prioritization• 2002:

– Continued focus on Aging Assets and Project Prioritization – Some work on Measuring & Valuing Reliability

• 2003:– Aging Assets – refine existing software & explore better ways to

analytically address the problem– Project Prioritization: Refine existing software & continue improve

methods for measuring T&D value

8 8

Utility Infrastructure Decision Making – The Focus of Our Work

Underlying Objective: Minimize the lifecycle costs of distribution infrastructure while meeting customer needs for reliability & power quality

Area Infrastructure Planning

Area Infrastructure Planning

Managing Aging Assets

Managing Aging Assets

Project Prioritization / Capital Budgeting

Project Prioritization / Capital Budgeting

Managing existing assetsManaging existing assets

Expand distribution infrastructure to meet future load

Expand distribution infrastructure to meet future load

Measuring & Valuing Reliability

Measuring & Valuing Reliability

Assessing Customer Needs

Assessing Customer Needs

Investment EconomicsInvestment Economics

Supporting Research

Decision Tools & Information

Utility Infrastructure Decisions

9 9

Our Focus

• Analytical Tools & Information– Managing Distribution Aging

Assets– Project Prioritization– Measuring & Valuing

Reliability – Assessing Customer Needs– Infrastructure Investment

Strategy

• Supporting research:– Measuring & Valuing

Reliability– Assessing Customer Needs– Investment Economics

10 10

Agenda



11 11

T&D Asset Management is Complex

• What about reliability - how far can we push the system before falling off the edge?

• How do we get the right-sized capacity when it is needed?

• Can we save $ by deferring maintenance?• What risks are we taking by practicing “doing-less”

decision making (do we really know how close we are to the edge)?

12 12

T&D Asset Management is Mostly About Solving Investment Problems

• 40% to 50% of electric utility net investment

• Business issues– Minimizing investment costs – Having “right” infrastructure to

meet customer needs– Making money

• Key strategic needs– Managing assets– Linking investment and O&M

decisions to customer needs

Net Invest.Gen. = $8.7B

Tran. = $4.5B

Dist. = $13.5B

Total = $26.7B

Generation Generation

A

B

230 KV

Backbone Transmission Grid

C

230/116

115/12 115/12

230/69

69/469/12

Distribution PlanningArea 1


GENERATIONPLANNING

TRANSMISSIONPLANNING

DISTRIBUTIONPLANNING

13 13

T&D Asset Management

• Two driving facts– Very low revenue to asset

ratio– Large embedded asset base

• Substantial care and feeding is required– Repair / Replace– Expand / Prepare for future– Reliability Net Invest.

Gen. = $8.7B

Tran. = $4.5B

Dist. = $13.5B

Total = $26.7B

Generation Generation

A

B

230 KV

Backbone Transmission Grid

C

230/116

115/12 115/12

230/69

69/469/12



GENERATIONPLANNING

TRANSMISSIONPLANNING

DISTRIBUTIONPLANNING

14 14

Opposing Trends are Changing the Business

• Increasing levels of key drivers– Restructuring & deregulation– Regulatory & Corporate

scrutiny – Customers demand for

reliability and service quality– Technology change

• Reduced budgets & increased corporate control

• Together these are creating a collection of problems

•Competition•Customer Needs•Technology

•Budgets•Local Control

Time

Leve

l

15 15

Nature of Business Planning Problem

Local AreaDecisions

Local AreaDecisions

TechnologyDemand for

Services

Regulators

ExistingAssets

CorporatePolicy

CorporatePolicy

Req’ts

Req’ts

Alternatives,Consequences

Budgets,Policy

16 16

T&D Asset Management – Project Planning & Financial Planning

• Fundamental engineering economic problems– Maintenance / Repair /

Replace– Capacity expansion– System Risk Assessment &

Mitigation

• Financial planning problems– 3 to 5 year Capital

Budgeting / Project Planning– Long-Term Financial

Planning

17 17

In Summary T&D presents special challenges

• Huge investment in assets• Regulations / Oversight• “Right” infrastructure for

customer needs• Competitive pressures - extract

maximum value from every asset• Performance vital to other

industries & economy– Recent NE power blackout estimated

to cost $7-10B– Loss of electric power to city of New

York costs about $36m/hr

18 18

Agenda

• Overview– Some Asset Management Beliefs: fact or fallacy?– T&D Asset Management Issues– History of our Research

• Purpose of Seminar– What do you need to know?– What do you hope to get out of the seminar?– What are the main asset management problems?– What have we missed?

• Using Models

19 19

Agenda



T&D Analytic Tools


May 5-7, 2004Hyatt RegencyBoston, Mass

Workshop on

Using ModelsLee MerkhoferLee Merkhofer Consulting22706 Medina CourtCupertino, CA 95014

408 446 4105www.prioritysystem.com

3c 2003, 2004 Lee Merkhofer Consulting

Fundamentally, good asset management is about making good decisions

“Asset management is a business process that allows a utility to make the right decisions on the acquisition, maintenance, operation, rehabilitation, and disposal of assets.”

(Stephen Kellogg, Vice President, CDM)


Fundamentally, good asset management is about making good decisions

• Asset management decisions are critical because:– The assets that are owned and how those assets are used strongly

affects the revenues that are generated.– The costs of acquiring, installing, maintaining and retiring assets are

typically large and directly affects the bottom line.– The capacity, reliability, and quality of the company’s assets strongly

affect customer satisfaction along with the perceived and actual levels of service.

“Asset management is a business process that allows a utility to make the right decisions on the acquisition, maintenance, operation, rehabilitation, and disposal of assets.”

(Stephen Kellogg, Vice President, CDM)


Numerous analytic tools are being promoted for asset management

Do such tools really help? What characteristics does a good tool need to have?What type of tool is best?

• Simulation models• Scoring tools• Executive dashboards• Narrow-focused applications• Enterprise-wide applications• Generic software• Custom software

Examples:


The quality of a decision tool depends on the quality of its decision model

“All models are wrong, but some are useful.”(George E. P. Box, University of Wisconsin)

AlternativesDecision

Model

Value

• Tools rely on analytic models.

• The model serves as a means for testing and evaluating alternatives.


More needs than resourcesHigh stakesChoices produce impacts over many yearsRisk and uncertaintyCompeting objectivesMultiple stakeholdersInterdependencies

Models are cost-effective for complicated decisions

Power Delivery Asset Management Decisions

Characteristics of Complex Decisions

YesYesYesYesYesYesYes

Etc.


Models are useful because they address key limitations of human problem solving

Empirical research shows people:- Have limited information processing skills- Can be biased- Can be inconsistent at making choices

…computers to do the things people have trouble doing.

…people to do the things that they are good at.

But people are good at:- Recognizing structure- Making “small,” well-defined judgments- Being creative

Models allow…


If well-designed, a decision model can provide 3 types of benefits1. Use limited resources more effectively by

– identifying value-maximizing choices– aligning decisions with corporate objectives– promoting better spending alternatives by clearly communicating performance

metrics– clarifying sources of value





2. Improve the decision-making process by – providing a framework for collecting and incorporating relevant information into

decision-making process– documenting performance goals and improving accountability– reducing opportunities for “gaming”– providing mechanisms for involving stakeholders in decision making– promoting consistent logic for valuing spending options and “leveling the playing

field”– providing discipline for making tough decisions– controlling the role of politics





2. Improve the decision-making process by – providing a framework for collecting and incorporating relevant information into

decision-making process– documenting performance goals and improving accountability– reducing opportunities for “gaming”– providing mechanisms for involving stakeholders in decision making– promoting consistent logic for valuing spending options and “leveling the playing

field”– providing discipline for making tough decisions– controlling the role of politics

3. Improve decision-making defensibility by– documenting underlying assumptions and decision logic– promoting consensus over spending decisions– answering “what if” questions


The dilemma: Complex decisions require sophisticated models

“There is a simple solution to every complex problem. Unfortunately, it is wrong.”

(H. L. Mencken)

An inadequate tool can mislead decision makers, resulting in poorer decisions than those that would be made without it.

How can users be assured that analytic tools are adequate decision aids? The workshop addresses this question.

T&D Project Prioritization

Lee MerkhoferLee Merkhofer Consulting22706 Medina CourtCupertino, CA 95014

408 446 4105www.prioritysystem.com


Topics• Project prioritization and T&D asset management• Introduction to priority systems• Mathematics of project prioritization• Valuing projects and project portfolios• Creating a priority system• Accounting for risk• Organizing for project portfolio management• Selecting project prioritization tools• Case study examples

– O&M resource allocation– Risk management– T&D project prioritization





Prioritization is an important concept for asset management

• Because resources are limited, not all asset investments can be made now– Prioritization is needed to determine which projects to do now, and

which should be postponed, scaled back, or eliminated.

Assets• Infrastructure• Property• Buildings• Equipment• Vehicles• People• Intellectual assets

Asset Investment “Projects”• Rehabilitate/repair• Replace/improve• Dispose• Create/acquire


Project prioritization guides asset management

• The goal of project prioritization is to select the set of projects that create maximum value.




• A “project” is any investment of organizational resources.




• A “project” is any investment of organizational resources.• The goal of asset management is to manage assets to

create maximum value.




• A “project” is any investment of organizational resources.• The goal of asset management is to manage assets to

create maximum value.• Thus, a priority system can be used to guide investments

for asset management.

PRIORITY LISTPROJECT CUM. COSTS

1. PROJ. W $1M2. PROJ. X $5M3. PROJ. K $8M4. PROJ. G $12M

27. PROJ. H $37M------------------------------------------28. PROJ. M $43M


Project prioritization fits the modern view of asset management

• Technical– Asset management is assuring high reliability and quality service

through adequately building and maintaining poles, wires, transformers, turbines, etc.

Old philosophy



• Technical– Asset management is building and maintaining poles, wires,

transformers, turbines, etc., to ensure high reliability and quality of service.

• Economic– Asset management is efficient capital rationing (using hurdle rates,

NPV, ROA, etc.).

Old philosophy

Recent philosophy



• Technical– Asset management is building and maintaining poles, wires,

transformers, turbines, etc., to ensure high reliability and quality of service.

• Economic– Asset management is capital rationing through financial evaluation

of options (hurdle rates, NPV, ROA, etc.).

• Strategic– Asset management is about obtaining maximum value from assets,

which requires integrating policy (strategic-level) information, systems-network understanding, and operations-management knowledge.

Old philosophy

Recent philosophy

New philosophy


The “paradigm” of modern asset management• Organizations exist because they create value

– For owners (who provide investment capital).– For customers (who purchase products and

services at a price they are willing to pay)




services at a price they are willing to pay)• The capacity to produce outputs of value to the

customer and, thereby, to create value for owners is directly related to the performance of the company’s production assets.




services at a price they are willing to pay)• The capacity to produce outputs of value to the

customer and, thereby, to create value for owners is directly related to the performance of the company’s production assets.

• Investments in the asset base (capital acquisition, operation, maintenance and renewal “projects”) should, therefore, be chosen with the goal of maximizing value.


Similarities between asset management and project prioritization

• “… should not just create value, it should create the greatest possible value.”

• “…requires seeing the value of managing assets from a strategic perspective.”

• “…is largely about information management.”

• “…involves seeking common solutions and practices across different lines of business and classes of asset types.

..seeks “value-maximizing”project portfolios.

…evaluates projects based on their impact on the fundamental objectives of the organization.

..uses rigorous techniques to obtain and analyze information.

..seeks accurate, logically defensible, and consistent solutions

Value-maximizing goal

Strategic perspective

Importance of information management

Search for consistent approaches/solutions

Asset Management Project Prioritization


However, formal project prioritization is a big step for most T&D companies

In most cases, companies:• Do not quantitatively evaluate and compare all distribution

projects.• The value of doing a particular project is not compared

with the values of competing projects. • Ignore uncertainty, project interdependencies, and the

strategic “option value” of investments.• Inadequately address process issues--including

information quality, credibility, and trust.





A priority system is a decision support tool

Rank Project7 ~~~~~8 ~~~~~9 ~~~~~

10 ~~~~~

Decision Support

scoresform

scoresform

TECHNICAL DATA

Software

Tota

l Ben

efit |

|||||||||

Total CostTotal BudgetTot

al B

enef

it

Data collection processes

Evaluationprocesses

Also known as tools for– Resource allocation– Multi-project management– Project portfolio management– Capital allocation– Enterprise asset management

Used to – Collect and consolidate project

data– Evaluate and compare project

alternatives– Provide decision

recommendations– Explain and defend choices– Create a level playing field for the

competition for resources


What is your biggest challenge?

Prioritizing which projects to do

Coordinating efforts across the company

Moving at the speed of the market

Finding qualified people

Insufficient resources

Convincing executives of importance

Identifying strategy

40%

36%

30%

22%

20%

20%

16%

Priority systems are popular because many managers view prioritization as a top challenge

(PricewaterhouseCoopers survey of managers, 10/00)


Project prioritization is increasingly being recognized as a key component of business success

“Of all the things a manager can do, figuring out the best way toallocate capital is probably the most important.”

(Fortune Magazine).

Choosing projects is of critical importance…




(Fortune Magazine).


…as is justifying the resources required.

“Regardless of industry, people are asking for more rigorous cases that illustrate both the short-term impact and the long-term benefits of projects.”

(Randy Hancock, Senior VP of strategy at Mainspring Inc.)




(Fortune Magazine).


…as is justifying the resources required.

Thus, project prioritizationis key.

“Overwhelmingly, enterprises that regularly prioritize projects and services by their business impact are increasing shareholder value.”

(Doug Lynn, MetaGroup)

“Regardless of industry, people are asking for more rigorous cases that illustrate both the short-term impact and the long-term benefits of projects.”

(Randy Hancock, Senior VP of strategy at Mainspring Inc.)


Priority systems address important questions for asset management

How should we spend our resources?

• What total budget should be sought from funding sources?• How should funds be allocated across organizational units?• What level of funding should be provided for each project?


Where should we be headed?

• Is the current set of projects the best way to accomplish our mission?

• To improve alignment with strategy, which projects should be accelerated and expanded, and which should be slowed, scaled back, or eliminated?

Priority systems address important questions for asset management (continued)


Priority systems address important questions for asset management (continued)

What is the right project mix?

• Balance between– Low-payoff “sure things” versus high-payoff “gambles”?– Near-term versus long-term payoffs?– Maintenance versus growth?– Risk versus return?


Continualimprovement

Identify needs

Eval

uate

pro

ject

s

& pr

ojec

t por

tfolio

s

Mon

itor p

rogr

ess

& ef

fect

iven

ess

Implementselected projects

Defineprojects

Priority systems are intended to support the on-going process of project portfolio management


Numerous needs,limited resources

Priority systems address key challenges

Competing objectives

Marketshare

Customer service

ReputationRegulatorycompliance

Financialperformance

Dispersed information/Multiple stakeholders/

Differing opinions

?

Risk/Uncertainty

D ecisions

Critical review

Biases/Gaming

Health, safety &environment


Project mix inconsistentwith strategy

≠Poor-performing portfolios

Projects that won’t die

Resources spread over too many projects

Over-emphasis on “low-hanging fruit”

…and overcome common problems of project portfolio management

R I S K

Insufficientunderstanding

of risk

“Fair-share” Allocations


The computational goal of a priority system is to identify the “efficient frontier” of project portfolios

• Concept: Some project portfolio (e.g., Portfolio D) are undeniably better than others.


The efficient frontier identifies the project portfolios that create the most value for the least cost


There are 4 basic types of priority systems

1. Project ranking systems2. Budget-allocation systems3. Combined (tiered) systems4. Problem/opportunity ranking systems


• Projects ranked according to benefit/cost• Typically used with an all-or-nothing, fund-from-the-top-down decision rule

• Produce a funding curve (efficient frontier) showing which activities to cut (add) and specific benefits lost (gained) if budget is decreased (increased).

Project ranking system

Fun

ded

PROJECT CUM. COSTS1. PROJ. W $1M2. PROJ. X $5M3. PROJ. K $8M4. PROJ. G $12M

27. PROJ. H $37M------------------------------------------28. PROJ. M $43M

Project costs and benefits evaluated

Projects ranked based on B/C

scoresform

scoresform

TECHNICAL DATA

Tota

l Ben

efit

Added costAdded value

Total value

Total cost

Organizational element funding

Ranking produces total benefit versus total cost curve (efficient frontier)


Advantages and disadvantages of project ranking systems

• Ranking activities is a relatively simple, easy to understand concept

• Enables the organization and project proponents to defend individual project choices

• Most appropriate for organization with centralized decision making

• Ranking logic can be extended to handle project interdependencies

• All-or-nothing funding is unrealistic—may need to define and evaluate alternative “versions” of activities (e.g., base-case and reduced scope).


Budget Allocation

Funds Unit A Unit B Unit C Unit D$15M 2 3 5 5$16M 3 3 5 5$17M 3 4 5 5$18M 3 4 6 5$19M 4 4 6 5$20M 4 4 6 6$21M 4 5 6 6$22M 5 5 6 6$23M 5 5 6 7$24M 5 5 6 8$25M 5 5 7 8$26M 5 6 7 8$27M 5 7 7 8

::

Budget allocation system• Provides an optimal allocation of a fixed budget among competing organizational

units • Each organizational unit proposes activities to be conducted under several

different, pre-specified funding cases (e.g., base-case funding, reduced funding, enhanced funding, etc.).

• Each funding case is evaluated (based on the activities to be funded).

Unit A (E-bus) Funding CasesFunding Case

Activity Min Reduced Base MaxB2B site $1M $2M $2M $2MWeb store $1M $1M $2M $2MKid’s lifestyle portal 0 0 $1M $1MRecruiting site 0 0 0 $1M

Total $2M $3M $4M $5M

Each unit prioritizes activities, indicates funding under alternative unit funding

cases, and scores funding cases

Scores used to determine optimal allocation of total funds

• “Bang for the buck” used to allocate funds across sites


Advantages and disadvantages of budget allocation systems

• Not as intuitive.• Allows organizational unit to retain authority for choosing

which activities to conduct—most appropriate for decentralized decision making

• Interdependencies among activities do not create a problem (since groups of activities are scored).

• Allows consideration of partial funding of activities (doesn’t assume all-or-nothing decisions).

• Doesn’t provide benefit estimates at the level of individual activities.


• Individual organizational units use a common ranking tool to evaluate and prioritize projects while upper organizational levels use a budget allocation system to allocate resource across the organizational units.

• The value-maximizing allocation matches slopes:

Combined (tiered) system

Integrated Budget Allocation/Activity Ranking System

Budget allocation system

Projectranking

tool

Project ranking

tool

Enterprise level

Individual organizational elements

Project ranking

tool

• Project prioritiesTotal benefit vs. total funding curves

• Funding case guidanceBudget allocation criteria/weights

Element 1 funding

Ben

efit

Cost

Element 1Element 1

Element 2 funding

Ben

efit

Cost

Element 2Element 2


Advantages and disadvantages

• Automatically provides the necessary activity ranking information in the form needed for the budget allocation system

• More complex than either of the individual systems• Usually results as an expansion from a project ranking or

budget allocation system


Fun

ded

PROBLEMS1. PROBLEM W2. PROBLEM J3. PROBLEM K4. PROBLEM G

27. PROBLEM. H

Problem/opportunity ranking systems• Rank problems or opportunities, not projects or investments• Tell you where you should be focusing your attention, not what you should

be doing• Simpler and less constraining than other approaches

– No inputs required on costs or effectiveness of investments– Defends decision makers without constraining them—decision makers decide how

much to spend on top-ranked items • Typically used as “screening” systems

Each problem/opportunity is scored

Scores used to rank problems/opportunities

scoresform

scoresform

TECHNICAL DATA





The project prioritization problem

Find the set of projects that• Produces the greatest possible value (benefit)• Without exceeding the available budget


The project prioritization problem

Find the set of projects that• Produces the greatest possible value (benefit)• Without exceeding the available budget

Mathematicians refer to this a the “knapsack problem”


Mathematical formulation of the basic project prioritization problem

N = number of project opportunities

xi =

bi = benefit obtained if i’th project is fundedC = total available budget

0 if the i’th project is not funded

1 if the i’th project is funded

Maximize Σ bixi

Subject to Σ cixi < C

i = 1

i = 1

N

N

This is known as a “zero-one integer programming problem”


Prioritization is surprisingly difficult to solve mathematically

• The problem is the number of alternative project combinations.– If there are N possible projects, there are 2N possible project

portfolios– E.g., with 30 projects there are roughly 1 billion alternative project

portfolios!

• Most software packages for integer programming use the “branch and bound” solution technique.

• The solution is time consuming, even with modern computers.

• If there are more than about 200 possible projects, even the fastest computers may not be able to find the solution.

• Thus, most priority systems do not use branch-and-bound


Fortunately, a theorem from economics often applies • Suppose the projects are independent

– Any combination of projects can be selected– Neither the costs nor benefits of any project depend on what other projects are funded


Cumulative cost

Cumulative benefit

B/CBenefitCostProjectRank

2371.50.524M71969.5111J61868.5263H51562.53217G4841.54.54.51K37375255X22126122w1



• Then, ranking projects by the ratio of benefit-to-cost and funding from the top down approximately gives the value-maximizing project portfolio




• Then, ranking projects by the ratio of benefit-to-cost and funding from the top down approximately gives the value-maximizing project portfolio

Cumulative cost

Cumulative benefit

B/CBenefitCostProjectRank

2371.50.524M71969.5111J61868.5263H51562.53217G4841.54.54.51K37375255X22126122w1

Budget constraint:

$20 million

Funded projec

ts


The project ranking shows the order in which projects are added

Cum

ulat

ive

bene

fit

Cumulative cost

0.5123

4.556

B/C

71.569.568.562.541.53712

Cumulative benefit

7654321

Rank

MJHGKXw

Project

23191815872

Cumulative cost

$20 M

5 10 15 2520


10

30

20

50

0

40

70

80

60

Budget constraint

…which can be plotted to show the “Efficient Frontier”


Important considerations

• the proper ranking metric is “bang-for-the-buck.”– Not “bang”– Not “alignment with strategy”– Not “balance”– Not “points”

Note that:




• ranking gives and approximate solution, not an exact one.

Note that:




• ranking gives and approximate solution, not an exact one.• ranking will give the wrong answer if there are

interdependencies

Note that:


Case study: Getting the concepts right is critical

Example—Suspiciously performing priority system

– Company was using a scoring approach recommended by consultants:

» Candidate projects were awarded points based on performance in several dimensions (financial reward, strategic fit, leverage, probability of success)

» Points were added and results used to rank projects

– Something was wrong» Large project nearly always ranked near the top


Case study: Getting the concepts right is critical

Example—Suspiciously performing priority system

– Company was using a scoring approach recommended by consultants:

» Candidate projects were awarded points based on performance in several dimensions (financial reward, strategic fit, leverage, probability of success)

» Points were added and results used to rank projects

– Something was wrong» Large project nearly always ranked near the top

– Approach had numerous “fatal flaws”» Most notably, failing to divide by cost (to get bang-for-the-buck) meant

that the logic was wrong. The system was biased toward large projects.





What exactly is “value”

“Asset management should not just create value, it should create the greatest possible value”

(Asset Management Software Vendor)


What exactly is “value”?

“Asset management should not just create value, it should create the greatest possible value”

(Asset Management Software Vendor)

• Such financial metrics fail to capture the true objectives of organizations

Two views on value:- shareholder value- stakeholder value

• Value is NOT – ROI, – ROA, – IRR, – NPV, – pay-back period, – etc.


Many US business leaders and management scientists believe the one and only goal of business is to create value for shareholders

“Maximizing shareholder value is becoming the number one priority for most publicly listed corporations.”

CFO, Price Waterhouse, 1997.





“Maximizing shareholder value is now embraced as the “politically correct” stance by corporate board members and top management in the United States.”

Alfred Rappaport, Creating Shareholder Value, 1998.



"Managing for shareholder value...pays off... We refined the goal over the years, constantly trying to come up with measures that better reflected the intrinsic value of the company andeach of the businesses. But we always maintained a single overall objective: generating greater value for the shareholder"

Brian Pitman, CEO of Lloyds Bank, HBR, April 2003.



“Maximizing shareholder value is now embraced as the “politically correct” stance by corporate board members and top management in the United States.”

Alfred Rappaport, Creating Shareholder Value, 1998.


Arguments for adopting shareholder value as the ultimate metric

• For publicly held companies, the value assigned by markets is, arguably, the final measure of business success




• Market valuations are objective and applied consistently across all publicly held companies and markets– Assured by principle of no arbitrage




• Market valuations are objective and applied consistently across all publicly held companies and markets– Assured by principle of no arbitrage

• Companies not managing for shareholder value are candidates for investor takeover


• Shareholder value is NOT the risk-adjusted, discounted value of projected future cash flows.

Managing for shareholder value and managing for profitable cash flows are two, very different things




• Example (4 power delivery companies):

Gap (=“option value”)

NPV of projected earnings (discounted at WACC)




• Example (4 power delivery companies):

• What causes these differences?

Gap (=“option value”)

NPV of projected earnings (discounted at WACC)


Stakeholder Value represents the alternative view

“It is probably safe to say there remains, worldwide, a large segment of the people that highly doubts the business goal of maximizing shareholder value is aligned with the interests of society at large.”

Bartley Madden, CFROI Valuation, 2000.


Stakeholder Value represents the alternative view

“It is probably safe to say there remains, worldwide, a large segment of the people that highly doubts the business goal of maximizing shareholder value is aligned with the interests of society at large.”

Bartley Madden, CFROI Valuation, 2000.

The alternative view is that businesses have obligations to other stakeholders as well:

• customers• employees• the community• suppliers• business partners• etc.

Company leaders must decide whether stakeholder values matter and what weight (if any) to assign to each stakeholder.


What kind of metrics reflect impact on value?

• Most organizations take the wrong approach. They measure what is easy to measure, not what is important.




• They take a bottom-up approach—they define interesting metrics, but then can’t come up with algorithms for computing value based on those metrics.




• They take a bottom-up approach—they define interesting metrics, but then can’t come up with algorithms for computing value based on those metrics.

• Unless there is a way to logically combine the metrics to determine the value added by projects, the metrics will not be of much use for prioritizing projects.


Use “value modeling” to define metrics

Value modeling• Subject of numerous books and articles, taught

any many graduate school business and engineering programs.

• Also called− “multiattribute utility analysis”− “multicriteria decision making”− “multi-objective decision analysis”

• Goal is to construct the most accurate and objective measure of value possible

System Performance

System Performance

SafetySafety PowerQualityPowerQuality ReliabilityReliability

Net RevenueNet Revenue EnvironmentEnvironment





The philosophy underlying the decision analysis approach to prioritization

• Resources are insufficient to do everything at once.• Things that are most valuable and most urgent should be done

first.• The value of an investment or other alternative is determined by

the degree to which it contributes to the achievement of objectives.

• Objectives and the weights to be assigned to them should be established by policy makers.

• In the absence of hard data and reliable models, estimates of the ability of alternatives to achieve objectives should be providedby those most knowledgeable about the alternatives and the needs that they will address.

• Techniques should be used to reduce biases and discourage gaming.


Basic steps for developing a priority system

1. Identify and structure the objectives– Specify what you want

2. Define performance measures– Specify how you will measure the degree to which an

alternative would achieve each objective

3. Derive the aggregation equation and decision rule– Specify how you will combine the performance measures

– Specify relative importance or weights

– Equation may be linear or non-linear and is derived using value modelling principles

– Define how projects will be prioritized


A simple example

What to pack for vacation?

AlternativesLimited resource: space in suitcase(5,000 cubic inches)


Structuring objectives as a hierarchy shows how achieving lower-level objectives (that are impacted by choices) enable higher-level objectives to be achieved.

Step 1: Identify and structure objectives

Avoid bringing heavy items

Avoid bringing big items

Bring items whose use enhances enjoyment

Bring items likely to be used

Maximize enjoyment of

vacation

Maximize fun Minimize work


Step 2: Define performance measures

Objectives Performance Measures Required Estimates

Avoid bringing items big items

Avoid bringing heavy items

Bring items whose use enhances enjoyment

Bring items likely to be used

Estimated volume (length x width x height, in inches)

Estimated weight in pounds

Benefit score:

Estimated number of times item would be used

Item size

Item weight

Benefit per use

Frequency of use

4 = Major benefit added3 = Significant benefit added 2 = Moderate benefit added

(about ½ that judged “major”)1 = Minor benefit added0 = No benefit added


Step 3: Derive the aggregation equation and corresponding decision ruleExample equation:

Ranking Measure = [ 80 x (Frequency of Use x Benefit per Use) - 20 x Weight ] / Size


Desired characteristics

• Logically sound (defensible) and accurate• Complete


Project benefit = Value (with project) –Value (without project)

OR

Project benefit = Value (project done now)-Value (project delayed)

Wn$ W1

1. Structure objectives

3. Estimate performance using data, models & judgment

4. Assess weights and other value parameters (scaling)

5. Combine to obtain total project value

Aggregationequation

Maximizeachievement of

organizational objectives

Maximizeshareholder

value

Maximize quality of organizational

image

Community

Maximize stakeholdervalue

Customers Workers

1

High

Low

5

1

High

Low

5

⌧

Value

Minimize adverse regulatory impacts

Overview of the logic for evaluating projects using a value model

2. Develop performance models/scales for lowest- level sub-objectives

2000 2001 2002 2003 2004Summary Income StatementPremiums $22,695.2 $23,998.8Net Investment Income 681.0 693.9Fees and Other Income 2,045.4 1,977.5

Total Rev. Excluding Cap. Gains $25,421.6 $26,670.2 $28,003.7 $29,403.9 $30,874.1annual growth rate 22.8% 4.9% 5.0% 5.0% 5.0%

Hard Benefit Incremental Revenue, Portfolio 0.0 0.0 0.0 0.0 0.0Hard Benefit Incremental Revenue, Single Initiative 0.0 0.0 0.0 0.0 0.0

Total Revenue $25,421.6 $26,670.2 $28,003.7 $29,403.9 $30,874.1

Current and Future Benefits (COGS) 19,340.2 20,433.4 21,562.9 22,641.0 23,773.0Gross Margin $6,081.4 $6,236.8 $6,440.9 $6,762.9 $7,101.0

percent of revenues 23.9% 23.4% 23.0% 23.0% 23.0%

W2

Financial model Scoring scales

Tota

l Ben

efit |

|||||||||


al B

enef

it

Performance models

"Hard" financialvalue

Platform forfuture success

Maximizeshareholder

value

Sub-objectives

Customers Workers Community

Maximizestakeholder

value

Minimize adverseregulatory impacts

Sub-objectives

Maximize qualityof organizational

image

Maximizevalue


Step 1: Structuring objectives

• The objectives hierarchy should consist of fundamental (not means) objectives.

For the value model to be logically sound, the objectives must satisfy certain rules…




• To ensure a simple aggregation equation (weight and add), the objectives should be “preferentially independent”– the importance of achieving any objective must not depend on the level of

performance achieved on any objective at the same level of the hierarchy

– this facilitates the evaluation of alternatives by ensuring that the weights specifying the importance of objectives will be constants





• To ensure a simple aggregation equation (weight and add), the objectives should be “preferentially independent”– the importance of achieving any objective must not depend on the level of

performance achieved on any objective at the same level of the hierarchy

– this facilitates the evaluation of alternatives by ensuring that the weights specifying the importance of objectives will be constants

• Lower-level objectives that are not preferentially independent cannot be weighted. Equations/models and other methods must be used to relate performance on these objectives to performanceon the higher-level objectives.



Case study: One T&D company’s view of value

• The first step was to get senior executives to agree on “value to whom?”

Example – Defining stakeholder value

Maximize total stakeholder value

Achievecompany financial

objectives

Provide quality,Low-cost service to

customers

Improve localcommunityperceptions

Reducecosts

Increaserevenue

Meet customer

commitments

Provide reliableservice

Charge reasonable

price

Timely & Accurate outagecommunicationShareholders Customers Citizens


Case study: One T&D company’s view of value

• The first step was to get senior executives to agree on “value to whom?”• The second step was to identify sub-objectives

Example – Defining stakeholder value

Maximize total stakeholder value

Achievecompany financial

objectives

Provide quality,Low-cost service to

customers

Improve localcommunityperceptions

Reducecosts

Increaserevenue

Meet customer

commitments

Provide reliableservice

Charge reasonable

price

Timely & Accurate outagecommunication


Sample objectives hierarchy developed for a county water utility

Provide healthy & & safe environment

and enhancedquality of life

Protect health &safety

Protectthe

Environment

Satisfy current& future

water needs

Prevent adverseeconomicimpacts tocommunity

Providerecreationalopportunities


Step 2: Defining metrics for measuring the degree to which objectives are achieved

• Performance measures quantify the degree to which objectives are achieved

• Performance measures should be “observables”– It’s 5 years in the future. What data could you collect to assess

performance against objectives?

• Choose measures that span the space of possibilities– Capture the attributes of what you expect to happen, and what might

happen.


• Objective: Customer satisfaction• Sub-objective: Minimize customer power outages• Performance measure: Outage duration• Possible events: Transformer trip, transformer fire,

explosion, fallen wire,…• Minimum duration--0 hours• Maximum duration--several days• Performance measure is a constructed scale based on

natural units.• Define scale using natural units.

Scale: 0 1 2 3

Example—Outage duration


Influence diagrams may be used to identify measures for creating constructed scoring scalesProcess for using influence diagrams:

1. Identify factors that influence achievement of the objective.2. Determine the relationship and relative importance among key factors

*Opportunity

for newservices &markets

Negotiatingleverage

Timeto market

Strategiccompetitivepositioning

In-sourcing/outsourcingpartnership

opportunities

*Technology

Ability totailor

Scaleability

Cable ofevolving

(adaptability)

Connectivity

Security

Electronicrecords

Abilityto conduct

surveysInformationrelevance

Informationscope (compre-hensiveness)

Robustgiven futureuncertainties

*Positive

visibility of actions tomarketplace

Info forunderstanding

value chain

Sub-objective:Provide

platform for future success

Illustrative


Based on value drivers in the influence diagrams, scales are developed for measuring each type of soft project benefit

*Opportunity

for newservices &markets

Negotiatingleverage

Timeto market

Strategiccompetitivepositioning

In-sourcing/outsourcingpartnership

opportunities

*Technology

Ability totailor

Scaleability

Cable ofevolving

(adaptability)

Connectivity

Security

Electronicrecords

Abilityto conduct

surveysInformationrelevance

Informationscope (compre-hensiveness)

Robustgiven futureuncertainties

*Positive

visibility of actions tomarketplace

To qualify for a score of 3, the initiatives must produce a significant positive impact on the company’s platform for success by allowing opportunities to provide new services or enter new markets and/or increase negotiating leverage, decrease time-to-market, facilitate profitable insourcingand outsourcing arrangements, and improve competitive position. This conclusion is based on a judgement that the initiatives will do one of the following(indicate which):1 Provide a significant new technology that is readily tailorable, scaleable, robust given future uncertainties, adaptable, and secure. For example, the technology will provide the greatest possible secure connectivity with all constituency groups, or2 Provide significant new and relevant information that will substantially improve understanding of the company value chain, or.3 Provide a clear signal to Wall Street that will be interpreted by analysts as evidence that company is exceeding expectations.

Info forunderstanding

value chain

Sub-objective:Provide

platform for future success

5

4

3

2

1

Scoring scale & sample definition


Portion of sample value model developed for T&D company


Step 3: Use data/models/best judgment to evaluate performance with and without each projectBasic logicKnowledge of…– Customers and markets– Levels of service provided (target and actual)– Existing assets and performance characteristics




Total Revenue $25,421.6 $26,670.2 $28,003.7 $29,403.9 $30,874.1



Financial model

1

High

Low

5Scoring scales

Tota

l Ben

efit |

|||||||||


al B

enef

it

Performance models


Step 3: Use data/models/best judgment to evaluate performance with and without each projectBasic logicKnowledge of…– Customers and markets– Levels of service provided (target and actual)– Existing assets and performance characteristicsProvides us the ability to predict…– Demand– How and when assets will fail– The consequences of failures




Total Revenue $25,421.6 $26,670.2 $28,003.7 $29,403.9 $30,874.1



Financial model

1

High

Low

5Scoring scales

Tota

l Ben

efit |

|||||||||


al B

enef

it

Performance models


Step 3: Use data/models/best judgment to evaluate performance with and without each projectBasic logicKnowledge of…– Customers and markets– Levels of service provided (target and actual)– Existing assets and performance characteristicsProvides us the ability to predict…– Demand– How and when assets will fail– The consequences of failuresWhich allows us to…– Evaluate alternative asset acquisitions, operations, maintenance, and

renewal options– Prioritize based on benefit/cost– Optimize the allocation of resources




Total Revenue $25,421.6 $26,670.2 $28,003.7 $29,403.9 $30,874.1



Financial model

1

High

Low

5Scoring scales

Tota

l Ben

efit |

|||||||||


al B

enef

it

Performance models


Step 3: Use data/models/best judgment to evaluate performance with and without each projectBasic logicKnowledge of…– Customers and markets– Levels of service provided (target and actual)– Existing assets and performance characteristicsProvides us the ability to predict…– Demand– How and when assets will fail– The consequences of failuresWhich allows us to…– Evaluate alternative asset acquisitions, operations, maintenance, and

renewal options– Prioritize based on benefit/cost– Optimize the allocation of resourcesWhich, in turn, enables us to…– Maximize performance at the system level (and, thereby, maximize

value) and explain and defend choices




Total Revenue $25,421.6 $26,670.2 $28,003.7 $29,403.9 $30,874.1



Financial model

1

High

Low

5Scoring scales

Tota

l Ben

efit |

|||||||||


al B

enef

it

Performance models


Step 4: Assign scaling functions and weights to quantify value

• Scaling may be needed to specify the relative value of changes.– How good/bad is a score of 2 compared to a score of 1?

• Scaling can be accommodated by either– Applying a scaling function to the natural scale or to the uniform,

constructed scale– Making the scale non-uniform.

• Example: Project designed to reduce outage duration.


Scaling – Example

0 1 2 3Duration (days)

Is an outage of 3 days 3 times worse than an outage of 24 hours?

Is an outage of 1 hour 6 times worse than an outage of 10 minutes?


Scaling – Example

NaturalscaleNone10 min.1 hr.2 hr.4 hr.8 hr.16 hr.24 hr.3 days> 3 days

0 1 2 3Duration (days)

0 -100-20-15-7-3 -90

Value (0-to100 scale)

Like weights, scaling requires value judgments.

Outage duration (days)

Value

1 2 3

-100

-50

Non-uniform value scale

0-0.1-.6

-1.2-3-7

-15-20-90

-100

Scaling function


A quick way to assign weights

1. Decide which objective is “most important”Suppose a series of “ideal” alternatives existed, each of which would produce a “best possible” impact on one (and only one) objective. All other things equal, which would you choose?

2. Rank the objectives

3. Each participant (individually) assigns weights

4. Comparison of individual weights & discussion

5. Final group weight assignments


The simple approach involves assigning weights that represent the value of going from “worst” to “best”

The weight, w, represents what 100% of the scale is worth

0

0.5

1

Worst

Best

25%

Value added by hypothetical project = 25% x w

Hypothetical project that scores a 0.25

Worst performance we could expect

Best impact we could expect from projects

(Normalized) scores


Weight assignment can be facilitated using poker chips• Assign chips (weights) in proportion to the value of each

impact

• If one impact is judged twice as important as another, assign twice as many chips (etc.).

• Maximum of 100 chips.

• Assume an impact in one area produces no (positive or negative) impact in any other area.

• Number of chips on each objective corresponds to weights.


Step 5: Determine the form of the aggregation equation• Weight and add is only correct if objectives are preferentially

independent (need to check)

• Sometimes, performance measures should be multiplied• E.g., multiplying the likelihood something happens by the outcome if it happens.

• Appropriate ranking metric (if projects are independent and don’t differ much in their “window of opportunity”) is benefit (net value excluding budget year cost) divided by budget year cost

• If projects have limited windows of opportunities, useful ranking metric is dollar loss from delay divided by budget year cost

Project benefit = Value (with project) – Value (without project)

Loss from delay = Value (if project done this period) – Value (if project done next period)

Wn$ W1

Aggregationequation

Value

W2


Include example here

• Include detail on weight assessment





Are risk analysts heroes?

• According to a 1997 Business Week review of Peter Bernstein’s book Against the Gods: The Remarkable Story of Risk:

“Statisticians, in the telling of Peter L. Bernstein, arenothing less than Promethean heroes…people whomastered the calculation of probabilities,…and stolefrom the gods something more precious than fire--namely, the understanding of risk.”


The key to addressing risk is “embracing uncertainty”

• Voltaire: “Doubt is not a pleasant condition, but certainty is an absurd one.”

• Francis Bacon: “If a man will begin with certainties, he shall end in doubts; but if he will be content to begin with doubts he shall end in certainties.”


Definition of risk

• Risk A situation wherein different things might happen, and at least one of the possibilities is bad.

• Note– Risk involves uncertainty and value– Risk is not “probability” or “probability times consequence”. These are

measures of risk.

Discrete Risk Continuous Risk

Consequence if risk event occurs = $1M loss

Probability of risk event = 1 in 10,000


Embracing uncertainty requires:

1. Characterizing the risk

2. Adjusting value based on organizational risk tolerance


The first step for characterizing risks is to identify all of the types of potential adverse consequences that are of concern

Types of riskTypes of risk

Publichealth &safety

Publichealth &safety

Workerhealth &safety

Workerhealth &safety

Environmentalresources

Environmentalresources Regulatory

compliance

Regulatorycompliance Corporate

image

Corporateimage Financial

performance

Financialperformance Employee

commitment

Employeecommitment Customer

satisfaction

Customersatisfaction

Risks can adversely impact the achievement of any and all objectives.


The next step is identify the sources of risk

• What are things that, if they happen could lead to the adverse consequences of concern?

• Some sources of risk are internal to the project:– Technology risk– Implementation risk– Schedule risk– Cost risk

• Some sources of risk are external to the project:– Governance risks (e.g., Enron)– Market risks (e.g., the economy)– Weather risks– Legal/regulatory risks

Project risks are often independent (uncorrelated) across projects; the project portfolio reduces the importance of these risks through diversification.

External risk can affect many or all of the projects simultaneously, making them potentially more important.


A simple way to address project risks is to use consequence and likelihood scales

Example (public health risk)

Very unlikely (1 chance in 10,000 or less)

.0001

Unlikely (1 chance in 1000)

.001

Possible (1 chance in 100)

.01

Reasonably likely (1 chance in 10)

.1

Certain or almost certain

1

Probability

No health impacts would occur.0

Only small amounts of minor hazards are involved, or exposures will be highly diluted. Any health impacts that could occur will have minor consequences (e.g., discomfort) and be short term.

.1

Exposures to moderate hazards would occur. Any health impacts that occur would have moderate consequences (e.g., illness) and be temporary.

1

Exposures to significant hazards would occur. Health effect could result in extended hospital stay.

10

If the event happens, the health impact would be most serious (e.g., death, cancer, paralysis), due to exposures to extremely dangerous hazards (large amounts of deadly substances)

100

Consequence


If uncertainties are estimated, they can be propagated through models to quantify the uncertainty over value

Project Performance Risks External Uncertainties

Scenario 1Prob P-1

Scenario 2Prob P-2

Scenario 3Prob P-3

Project A PerformanceProb

abili

ty D

ensi

ty

Projects

•Project A

•Project B

•Project N

Composite Risk Model

Risk-SourceModel

RiskPropogation

ModelConsequence

Model

Prob

abili

ty D

ensi

ty

Number of Health Effects

Portfolio Risk Estimates


Quantifying uncertainty allows confidence bounds to be placed on performance forecasts


Sophisticated methods are available for quantifying risk tolerance

• However, unless project involve “bet the company” gambles, it is usually not necessary to use sophisticated techniques.


A simple (although imperfect) way to account for risk tolerance is through hurdle rates

• The hurdle rate is a risk-adjusted cost of capital used to discount future project costs and benefits

• Higher hurdle rates are applied to projects considered to be more risky

25%High risks (e.g., untested technology)WACCAverage risks

6%Below average risksRisk-free rateVirtually no risk

Project Risk Hurdle Rate

EXAMPLE

• Hurdle rates are a “quick-and-dirty” way of addressing project risk.


Risk and project prioritization

• Characterizing risk is relatively easy (and useful)– E.g., rather than obtain a “point estimate” for some uncertain aspect of

project performance, obtain a range.– If there are rare, but high consequence risks, estimate performance

contingent on the occurrence of the risk, plus a probability.– Be careful of external risks and other risks that may affect several or all

projects simultaneously. They can create portfolio risks.

• Characterizing risk tolerance is usually not that important– Exception is “bet the company” projects, for which methods more

sophisticated than hurdle rates should be used.





Organizing for project prioritization requires recognizing the importance of managing at the portfolio level

• Most organizations put ample effort into making individual projects successful– Highly motivated project managers

– Once funding obtained, focus is on achieving the costs, schedules, and performance mandates of their specific projects

• …but put insufficient effort into making the entire portfolio of projects as successful as it could be– Either no one has responsibility for the overall portfolio, or

management of the project portfolio is not as effective as it could be.


Problems with project-by-project decision making

• In most companies, the basis of a “go” decision is whether the project passes some hurdle

– Many projects pass the hurdle

– ..but, resources are highly constrained

– ..so people become more than 100% committed

– …which leads to the downward spiral of poor project portfolios.


Need to establish a project portfolio management office• Project portfolio manager has accountability for the

success of the entire project portfolio– PPM given an estimate of funding available– Up to PPM to decide how to allocate the funds across projects (at

minimum, PPM makes recommendations for final approval by executive committee)





• The portfolio manager is supported by a portfolio management team






• Team is responsible for – Evaluating project proposals– Validating project cost and performance estimates– Accepting/rejecting proposals– Accelerating and decelerating projects– Allocating resources– Continually managing the project portfolio






• Team is responsible for – Evaluating project proposals– Validating project cost and performance estimates– Accepting/rejecting proposals– Accelerating and decelerating projects– Allocating resources– Continually managing the project portfolio

• One member of the team is appointed primary contact for each project manager


Typical project portfolio management process





Numerous analytic tools are being promoted for asset management

Do such tools really help? What characteristics does a good tool need to have?What type of tool is best?

• Simulation models• Scoring tools• Executive dashboards• Narrow-focused applications• Enterprise-wide applications• Generic software• Custom software

Examples:According to META Group estimates, the project portfolio management tool market was about $85 million in 2002; by 2005 it could reach $540 million.


To evaluate a tool, consider how well it performs against 6 criteria

1. Accuracy2. Logical soundness3. Completeness4. Practicality5. Effectiveness6. Acceptability

The remainder of this section will clarify these considerations and provide observations, examples, and advice.


The tool must be accurate

• Does the tool:– Produce the “right” answer?– Provide outputs within an acceptable confidence level and precision?– Indicate the confidence or uncertainty associated with outputs?

• Is the tool biased toward or against certain projects, interests, or considerations?

• Are results highly sensitive to untested or untestable assumptions?


Observation: Intuitive and well-established models are not necessarily accurate

Example - Models for Ranking Superfund Sites

• Congress required the EPA to evaluate whether its Hazard Ranking System (used to place sites on the National Priorities List) was the best-available site-ranking model.

• The analysis consisted of comparing site rankings produced by various models with a ranking produced by an independent expert panel.

• Results showed none of the models produced rankings that correlated with each other or with the ranking provided by the expert panel (Call and Merkhofer 1988).

If a tool is logically sound and complete, its recommendations are more likely to be accurate.


The tool must be logically soundConsiderations for assessing logical soundness

• Degree to which tool can be justified (in terms of relevant theories and empirical evidence)

• Operational validity – Are there problems with satisfying underlying assumptions?

Note: Mathematical theories for valuing and prioritizing projects exist and are well-accepted within the technical and academic communities (e.g., multi-attribute utility analysis, portfolio theory, real options, etc.).


Observation: Tools not based on the relevant theories are time bombs—they go off when subjected to technical review Example – DOE’s System for Ranking Sites for the Nuclear Waste Repository

• Hanford

• Yucca Mtn

• Deaf Smith County

• Davis Canyon

• Richton Salt Dome

• The DOE initially used a scorecard approach to rank sites being considered as locations for the repository.

• Hanford, Washington, was the top-ranked site.

• Responding to critics, DOE had the National Academy of Sciences (NAS) review the analysis.

• The NAS concluded that DOE’s method was “unsatisfactory, inadequate, undocumented, and biased.”

• DOE was forced to redo the analysis using an academically “correct” methodology (multi-attribute utility analysis).

• The result caused DOE embarrassment and forced it to revise its choice to Yucca Mountain, Nevada. (Merkhofer and Keeney 1987).


The tool must be complete, accounting for allrelevant decision considerations

• Does the tool ignore important considerations?• Is it possible to expand the tool to include omitted

considerations, or are the omissions inherent in the approach?

An enterprise-level view of asset management demands capability to evaluate impacts on all enterprise objectives.


Observation: Completeness requires capability to handle “hard” and “soft” considerations

Example – Prioritizing Investments in a Water UtilityTo fully value investments, it was necessary to

- Account for “soft” benefits, such as public health and safety, reduced flood risk, improved customer perceptions, community recreational opportunities, and increased organizational knowledge and capability.

- Integrate information from numerous databases, including customer information systems, GIS, work management systems, investment modeling applications, etc.

- Include scoring methods to capture local knowledge possessed by field personnel but not captured by available data.

The need for accuracy, logical soundness, and completeness is why asset management tools must be sophisticated.


The tool must be practical to implement and use

Do you have:- Necessary expertise to apply the tool?- Data required for input?- Computation resources for applications?- Sufficient time for applications (including time to collect input data)?

A utility can make it practical to use more sophisticated tools through training, reassigning responsibilities, developing new sources of data, and adjusting budget schedules.


Observation: Experience creates capability and demand for more sophisticated tools

• DOT regulations required client to adopt formal system for identifying risks and prioritizing risk-reducing activities.

• Implemented in 1995, the system was initially perceived by many as “overly complex.”

• However, each year the client expanded the system, making it more complete and sophisticated.

• Experience, better documentation, and internal training built confidence and expertise, promoting desire for more capability and accuracy.

Example - Pipeline Risk Management System

Avoid simplistic tools and tools whose capabilities cannot be expanded as experience and understanding grows.


The tool must be effective at achieving its intended goals

How will the tool be used:- As a backroom decision aid?- As a vehicle for creating internal consensus?- As a vehicle for persuading external parties?

What does the tool need to do well to address limitations of current practices?


Observation: Sometimes, being effective requires a new approach

Example—Valuing Long-Term Energy Contracts

– Utility client needed way to compare short vs. long term (>10yr.) energy contracts.

– Standard valuation methods were overly sensitive to assumed discount rate.

– Delivered system uses real options analysis, a new technique that » quantifies the “option value” of contracts (an energy contract

provides an option to buy throughout the duration of the contract).» does not require risk-adjusted discount rates

Actual Market value

Value computed using traditional methods

Opt

ion

valu

e


The tool must be acceptable to decision makers and other stakeholders

Is the tool:- Compatible with existing processes/organizational structure?- Understandable and understood?

Stakeholders feel threatened by tools that impact funding. For success, they must have confidence that the tool will help them and the business succeed.

Unless all parties want it, understand it, and trust it, the tool will be abandoned.


Observation: Developing the tool as a collaborative effort promotes buy-in and allows the tool to be designed to fit the need

Example – Prioritizing Investments in Information Technology

• Client desired system for prioritizing proposals submitted by different departments.

• Core Systems Department convinced tool would not work, “How do you value investments in capacity?”

• Design gives infrastructure that enables other projects a portion of benefit generated by enabled projects.

• Tool encourages Core Systems to propose infrastructure that serves needs of other departments.

Case Study: A Project Prioritization System


Background

Motivation and objectives• Western electric and gas utility operates

roughly 3000 facilities in seismically active locations

• Asset risk manager concerned that senior management does not appreciate the risks and is not providing sufficient resources for risk management

The developed system• Screens facilities to identify those that warrant formal analysis• Ranks facilities based on total seismic risk• Quantifies the risk-reducing benefits of activities proposed to reduce

risk• Prioritizes activities based on benefits and costs• Provides estimates of total seismic risk and ability to reduce total risk

as a function of risk-management budget


• Projects ranked according to benefit/cost• Typically used with an all-or-nothing, fund-from-the-top-down decision rule

• Produce a funding curve showing which activities to cut (add) and specific benefits lost (gained) if budget is decreased (increased).

Project ranking system

Fun

ded

PROJECT CUM. COSTS1. PROJ. W $1M2. PROJ. X $5M3. PROJ. K $8M4. PROJ. G $12M

27. PROJ. H $37M------------------------------------------28. PROJ. M $43M

Project costs and benefits evaluated

Projects ranked based on B/C

scoresform

scoresform

TECHNICAL DATA

Tota

l Ben

efit


Total value

Total cost

Organizational element funding

Ranking produces total benefit versus total cost curve


The objectives hierarchy for the system was developed by the project team and approved by senior management


To assess risks…

• Seismic risk models are used to quantify ground-shaking scenarios

• Structure models are used create damage scenarios• Scoring scales used to convert damage scenarios into

impacts on objectives

1

High

Low

5

Scoring scales


Example - Scale for worker health and safety

How many worker fatalities do you expect to result if the facility experiences the indicated damage outcome? Average for changes in population over time.

Score

0. None

1. 1 chance in 1000 of one fatality2. 1 chance in 100 of one fatality

3. 1 chance in 10 of one fatality

4. One fatality

5. 10 fatalities

6. 100 fatalities


Example - Legal liability scale

Score0. No chance1. 1 chance in 10 or less2. Between 1 chance and 50%3. Between 50% chance and 90%4. Almost certain

If the damage outcome occurs, how likely is it that the company will be held legally liable for impacts attributed to facility damage?

Score0. No chance1. Less than $1M2. Between $1M and $10M3. Between $10M and $100M4. Between $100M and $1B5. More than $1B

What is the most likely total dollar impact to the company if it is held legally liable for impacts?


Example – System functionality

Score0. None1. Less than 10,0002. Between 10,000 and 100,0003. Between 100,000 and 1,000,0004. More than 1,000,000

If the seismic event and damage outcome to the facility occurs, how many customers will lose service?

Score0. None1. Less than an hour2. Between 1 and 6 hours3. Between 6 and 12 hours4. Between 12 hours and 1 day5. Between 1 day and 3 day6. More than 3 days

If the facility damage outcome occurs, what is the likely duration of the lost service that it will cause, or, if the facility is needed to return service, by how long will service return be delayed?


Weights were assigned by company executives

One incidence of 3-day power loss to 3000 residential customers

No outage

One chance in 100 of a public fatality

No risk


System software collects inputs and quantifies risk


Projects that reduce risk produce benefit


Projects are prioritized based on the ratio of benefit-to-cost


Summary

• The priority system demonstrated to senior management that seismic risks are serious.

• Enabled proposed investments to be prioritized based on risk-reduction benefit.

• Forced senior management to take ownership of the problem.

• Provided auditable documentation of efforts to address risk.

Case Study: A Budget Allocation System

Originally prepared by

Elsie Martin, Myers Martin Consulting, LLCLee Merkhofer, Lee Merkhofer Consulting

for the

EPRI Power Delivery Asset Management WorkshopJune 3-5, 2003, New York, NY


Case study materials previously presented in public forums

• Xcel entered the application in the Institute for Operations Research and the Management Sciences (INFORMS) 2000 competition for “Best Application of Decision Analysis”

– Awarded “runner up” status

– Material herein presented during judging of finalists

• This material presented at the EPRI Power Delivery Asset Management Workshop, June 3-5, 2003, New York, NY


Background: Application describes priority system used 1998-2000 by Northern States Power (NSP)

• NSP is a large, Midwestern electric utility, now part of Xcel Energy

• Goal was to help NSP (Electric) allocate its $250 million operating budget

• The priority system was used to help the Asset Management department allocate operating funds to diverse business areas responsible for

– Delivery (Transmission & Distribution)– Retail Services (Call Center, Billing, Marketing & Sales)


By 1998, it was clear that the existing budget allocation process was no longer adequate

• Lengthy discussions, eloquence always won

• Sub-optimization by territory was not consistent with new matrix structure

• Need to deliver more service without increasing costs

• Although NSP had experience using a ranking system for prioritizing capital expenditures, such an approach was determined inadequate for operating budget decisions


Challenge was to create a priority system that would:

• Identify and quantify allsignificant benefits of operating expenditures

• Provide a structured process with appropriate input from all stakeholders

• Level the playing field

• Align budgets with objectives

• Incorporate available data on performance & from surveys as well as best-professional judgment on expenditure effectiveness

• Show what is gained/lostif budget is increased/decreased

• Improve “bang for the buck”


Budget Allocation

Funds Unit A Unit B Unit C Unit D$15M 2 3 5 5$16M 3 3 5 5$17M 3 4 5 5$18M 3 4 6 5$19M 4 4 6 5$20M 4 4 6 6$21M 4 5 6 6$22M 5 5 6 6$23M 5 5 6 7$24M 5 5 6 8$25M 5 5 7 8$26M 5 6 7 8$27M 5 7 7 8

::

System based on a budget allocation design• Provides an optimal allocation of a fixed budget among competing organizational

units • Each organizational unit proposes activities to be conducted under several

different, pre-specified funding cases (e.g., base-case funding, reduced funding, enhanced funding, etc.).

• Each funding case is evaluated (based on the activities to be funded).

Unit A (E-bus) Funding CasesFunding Case

Activity Min Reduced Base MaxB2B site $1M $2M $2M $2MWeb store $1M $1M $2M $2MKid’s lifestyle portal 0 0 $1M $1MRecruiting site 0 0 0 $1M

Total $2M $3M $4M $5M

Each unit prioritizes activities, indicates funding under alternative unit funding

cases, and scores funding cases

Scores used to determine optimal allocation of total funds

• “Bang for the buck” used to allocate funds across sites


The system was designed to allocate the budget across 10 business areas

• Business areas, called “portfolios” addressed various transmission, delivery, and retail services.

• Alternative 5-year funding scenarios were evaluated for each portfolio.

• Each portfolio team proposed between 3 and 9 alternative funding cases (5-year scenarios).

• Each funding case is scored against 7 objectives.


Each portfolio team considered several alternative funding scenarios

• Case 1 = minimum funding (minimum funding required to meet legal obligations)

• Case 5= base case (five years at `99 actuals)

• Case 9 = maximum funding (unconstrained funding) all activitieswith benefits that justify costs

• Cases 2-4, 6-8 = intermediate funding scenarios ($1-2M max between cases)

• Portfolio team specified activities to be funded under each case, then “scored” cases to estimate benefits

Case 5

Activity 3

Activity 1Case Year 1 Year 2 Year 3 ...1 0 0 $15 $1M $1.2M $1.2M9 $7M $7M $7.1M

Activity 2

Case 1

Yr 1 2 3 4 5

Case 9


Policy makers created the objectives hierarchy for evaluating the cases

Better, Faster,Cheaper

Financial CustomerService

Platform forSuccess

MeetCommitments

& BuildRelationships

ReliabilityAccurate &

TimelyBilling

Impact onCost toOthers

Profitsfrom Value-

addedServices

OutageCommunication

SystemReliability


A model was developed for quantifying benefits

• Measures and scales (derived from influence diagrams) were created for each objective to “score” performance.

• Weights for objectives were assessed from senior NSP Electric executives.

• A mathematical function was derived to translate “scores” and “weights”into estimates of “utility,” a measure of the value of each portfolio.

• System software computed the costs and benefits of all possible combinations of funding levels.


Influence diagrams were constructed to help select performance measures

For each funding case, estimate:• frequency of sustained outages• frequency of momentary outages• number of customers with > 4

sustained outages• average duration of outages• measure of power quality

1. Influence diagrams identify factors influencing customer’s perception of system reliability

Example: System reliability objective

2. Key factors define measures and scales used to forecast performance

`99 actuals3. Data on recent

performance provided as benchmark for scoring

Systemreliability

Outages Powerquality

CustomertypeFrequency

Duration

Momentary

Sustained

# ofcustomersaffected

Frequency

Magnitudeof disruption


Portfolios (equivalent

dollars)

Portfolio cost

Case 1

Case 5

Case7Case 9

Example: Billing and Payment Portfolio

Scaling functions were developed to translate performance “scores” into estimates of dollar value

• Scaling functions (e.g., reliability function) derived from experts

• Weights (e.g., swing weight for system reliability) assessed from senior NSP Electric executives

• Results show for eachportfolio how fundingimpacts value added


System software computes the costs and benefits of all possible combinations of funding levels

Total (5-yr) cost (millions of dollars)10 20 30 40 50 60 70

Total value added

0

20

40

60

80

100

120

140

160

180

200

0

Efficient frontier of optimal allocations

*Hypothetical data

Each dot represents a specified funding level (case) for each portfolio


A quality process was used to collect inputs, interpret outputs, and improve the system

• Scoping sessions• One person in Asset

Management leads each portfolio team

• 3 meetings with all 50 participants

• 30-page instructions booklet

• Documented funding cases

• Very rigorous scoring sessions!

• Lesson learned


Outputs include:

• Benefit vs cost curves for each portfolio

• Total benefit vs total cost for the operating budget

• Table showing priority order for incrementing each portfolio’s funding level

• Sensitivity analysis

• Incremental benefits for each objective

• Out-year funding requirements for each department manager

• Database containing activities to be conducted and benefits anticipated for each portfolio for all funding cases


Sample Output: Benefit vs. cost for Line Clearance

0.000

0.010

0.020

0.030

0.040

0.050

0.060

0.070

0.080

0.090

0.100

Case 1 - Case 4 Case 4 - Case 5 Case 5 - Case 6 Case 6 - Case 9

Case Increments

Ben

efit

Impact on Costs to Others Future Profits Accurate & Timely BillingSystem Reliability Outage Communication Meet Commitments & Build RelationshipsPlatform for Success

2.40

2.45

2.50

2.55

2.60

2.65

2.70

2.75

$70,000 $80,000 $90,000 $100,000 $110,000 $120,000 $130,

Cost

Cum

ulat

ive

Util

ity


Total benefit vs. total cost: Optimal allocations (based on 5-year benefits & 5-year costs)

Benefit to Total Cost Curve for Optimized Allocation

0.00

1.00

2.00

3.00

4.00

5.00

6.00

$950,000 $1,000,000 $1,050,000 $1,100,000 $1,150,000 $1,200,000 $1,250,000 $1,300,000

Total Discounted 5-Year Cost ($000s)

Cum

ulat

ive

Util

ity $275,000 in 2001


Year 1 funding steps (based on 5-year B/C optimization) for a $275 million budget

Funding level

Portfolio Abbreviation Cases Incr. Utility

Incr. 2001 Costs

($000s)

Cumulative Costs for 2001

($000s) 1 SOS Case 5 ('99 Actuals) 0.17 462$ 218,502$ 2 C&I Case 2, Case 3 0.55 3,422$ 221,924$ 3 B&P Case 3 0.29 3,046$ 224,970$ 4 C&I Case 5 ('99 Actuals) 0.19 1,505$ 226,475$ 5 SOS Case 6 0.35 688$ 227,163$ 6 SOS Case 7 0.29 750$ 227,913$ 7 Dist Sust Case 3, Case 3.5 0.28 3,453$ 231,366$ 8 EP&S Case 2 - Protect, Case 4-Begin Growth, Case 5 ('99 Actuals) 0.41 7,234$ 238,600$ 9 R&O Case 5 ('99 Actuals) 0.05 622$ 239,222$ 10 B&P Case 5 ('99 Actuals) 0.16 2,556$ 241,778$ 11 R&O Case 5.1 RTS+X1, Case5.2 RTS+GIS1+X1, Case 6 GIS2+RTS+X2 0.78 2,695$ 244,473$ 12 Met Read Case 3 - Limited AES rollout 0.06 2,430$ 246,903$ 13 Met Serv Case 2 0.08 2,248$ 249,151$ 14 Met Read Case 5 - Scheduled AES rollout 0.05 2,400$ 251,551$ 15 Trans Case 5 ('99 Actuals) 0.23 5,398$ 256,949$ 16 Trans Case 6 0.14 1,254$ 258,203$ 17 Dist Sust Case 5 0.10 3,394$ 261,597$ 18 Met Serv Case 4 0.01 642$ 262,239$ 19 LC Case 4 0.07 10,600$ 272,839$ 20 Met Read Case 7 0.07 1,305$ 274,144$ 21 LC Case 5 ('99 Actuals) 0.09 8,003$ 282,147$ 22 LC Case 6 0.05 1,700$ 283,847$ 23 EP&S Case 6-Prepare Compete, Case 9-Full strategy 0.25 5,794$ 289,641$ 24 R&O Case 7 GIS3+TRBL1 0.06 1,718$ 291,359$ 25 Met Serv Case 5 ('99 Actuals), Case 9 0.03 2,192$ 293,551$ 26 Met Read Case 9 - Growth(Data Dream) 0.05 1,873$ 295,424$ 27 SOS Case 9 0.13 4,900$ 300,324$ 28 B&P Case 7 0.06 2,384$ 302,708$ 29 R&O Case 9 TRBL2 0.09 4,262$ 306,970$ 30 B&P Case 9 0.03 1,265$ 308,235$ 31 Dist Sust Case 6.9 0.04 1,900$ 310,135$


What the system did wellNSP managers & experts saw significant (and unanticipated) benefits

Improvedunderstanding

• “I gained an appreciation of the impact on others. If I get something, someone else doesn’t.”

• “I gained an understanding of consequences beyond just economics.”

• “We made improvements in getting information up to management.”

(1 of 4 slides)


Improved accountability and responsibility

because managers must document what

they expect to accomplish

• “We had people digging into costs, finding out what builds cost structure.”

• “We gained an understanding, tying costs to benefits and to impacts of not spending.”

• “Building funding cases promotes a better understanding of what it takes to run your business.”

(2 of 4 slides)


Improved performance due

to clearer objectives

• Managers redefined and shifted dollars among activities to get higher performance

• “People acknowledged that there are alternatives”

• “In developing funding cases, we got a chance to talk with colleagues in our own portfolio about what excellence would be”

(3 of 4 slides)


Provided a framework for evaluating funding

requests that is fair, consistent, and

promotes a complete and thorough

consideration of relevant issues

• “It is a place where there is dialogue that is internally consistent.”

• “It worked! We made the tradeoffs to meet our budget.”

• “Areas could go off to do their own pieces and they would still fit together.”

• “Initially it seemed too overwhelming and complex. But, when we got into it, it worked.”

(4 of 4 slides)


In interviews after 2000’s application, NSP managers & experts saw needs that must be addressed to improve applications

Need for more and better dataNeed for controls to prevent

gaming

Need to better link portfolios with department budgets and

capital budgets

• “Bulk transmission doesn’t have adequate reliability drivers. The current ones are for measuring load serving reliability.”

• “[Distribution Maintenance] cases were too big and too generic. If low-value work gets lumped with big value dollars, they go together.”

• “We are trying to do two things – portfolios and departmental budgets – that don’t sync up.”

• “Capital & Operating allocations are separate processes. We have no way to compare preventive maintenance and a capital expense.”


Why was system canceled rather than modified as in previous years?

• NSP merged to be come Xcel Energy– Xcel management wanted their own management system– They wanted a system that relies on analysis of historical data and

statistical risk and less on professional judgment of those closest to problem

– Focus was on merger activities


Conditions for success of any portfolio optimization tool

• Senior management must understand and sponsor it

• Department heads must be willing to give up some autonomy for good of the whole, trusting that tool is fair

– Company culture is important!

• Scope must be clear

T&D Project Prioritization Case Study Example

Steve ChapelCharles Feinstein

2

Problem DefinitionThe company does not currently quantitativelyevaluate and compare all distribution projects. (A formal, repeatable, and uniform approach for valuing projects does not currently exist.)The value of doing a particular project is not compared with the values of competing projects. For the projects that are evaluated, the company is not satisfied with the current procedures.

3

Scope of Project Prioritization Problem

Large number of projectsMultiple performance measuresProjects done for different reasonsAnalysis of uncertaintyRisk of deferralRespond to budget signals

4

Characteristics of Project Prioritization System

Level playing field for all projectsResolve differences of opinion rationally

Techniques for resolving differences of opinion and determining which differences matter

Defensible logic for peer reviewTransparent analysisCompleteness with respect to performance measures

Multiple performance measures for multiple objectives

Bias- and error-free

5

Characteristics of P2 System -continued

Practically applicable with respect to time and costCompatible with existing business practicesExplicit treatment of uncertaintyAbility to quantify what is lost from insufficient fundingSoftware to manage and compare large numbers of diverse activities – client/server database (Oracle, SQL Server)

6

P2 Value Measurement System

The system is:Multi-year Multi-attribute Value driven

Three key dimensionsObjectives of the project portfolio

minimizing or maximizing important, measurable aspects of system performance

Values capture relative importance of competing

objectives.

Project attributes describe how each project contributes to

attainment of objectives

System Performance

System Performance

Safety PowerQuality Reliability

Net Revenue Environment

7

Required I/O + Transformations

INPUTS

•Corporate budgets

•Projects + Alternatives

•Objectives

•Values

•Attributes

OUTPUTS

•Project Rankings

•Portfolio of projects

•Timing of projects

•Value of additional budget

•Value

•Risks

•Costs

TRANSFORMATION PROCESS

•Attribute + values + objectives

Benefits

•Projects + Alternatives

Budget requirements

•Benefits + budget req’ts

Portfolio

•∆ Budgets

∆ Portfolio

8

Overview of System Structure

Objective Specification

Value Specification

Attributes Specification

Project Definition

Project Analysis

Portfolio Design

Budget Specification

Portfolio

9

Natural Units – measure system performance

Power QualityReliabilityCosts / RevenueEnvironmentalSafety

10

Scales – measure value of change

Linear DollarsNo. CustomersEtc.

Non-linearFrequencyDurationNo MomentariesEtc.

0

0.2

0.4

0.6

0.8

1

$0 $10,000 $20,000 $30,000 $40,000 $50,000

Net Revenue

Scal

ed V

alue

0

0.2

0.4

0.6

0.8

1

0 5 10 15 20

Frequency of Sustained OutagesSc

aled

Val

ue

11

Weights – measure relative value of change

Project Value

Power Quality 75 Reliability 100 Safety 100 Financial 70Environment 90

Momentaries 70

Residential 75

Commercial 100

Industrial 100

Sustained 100

Residential 75

Commercial 100

Industrial 100

Employee 100Residential 50

Commercial 100

Industrial 100

Emergency Days 100

Normal Days 100

Public 100

Low Density 40

Medium Density 75

High Density 100

12

Attribute Structure

At a high level, the attributes are the same

The differ in how the high level attributes are measured – Lets add details

Project Value

Power Quality Reliability Safety FinancialEnvironment

13

Attribute Structure - DetailedProject Score

Power Quality Reliability Safety Change in Revenue

Problem - In One Month Momentaries

Problem - In > One Month

Sustained

Residential

Commercial

Industrial

Overload Non-Overload

Residential

Commercial

Industrial

Residential

Commercial

Industrial

14

P2 System

Administrator Set-upData EntryAnalysis / Project Selection

15

Software Demonstration

Repair and Replace Strategy for Aging Distribution Systems Assets

Charles D. Feinstein and Peter A. MorrisVMN Group

andSteve Chapel

S. Chapel Associates

May 2004

2

Outline

• The Aging Asset problem• Substation transformer example—single asset• The need for a Learning Model – group exercise• Aging Asset model structure-single asset• Control of Asset Population - Cable Case Study• Organizational Issues

3

The Aging Asset Problem• Given

– An asset type (e.g., transformers, cables, poles, etc.)– A set of asset characteristics

AgeConditionFailure modesUncertainties in future performanceObservables and UnobservablesCosts

– A set of alternativesRepairReplaceRebuildRefurbishTestMaintain

• What should we do, when, and under what conditions?

4

Problem Statement

• The current system contains (x miles of cable, y power transformers, z breakers, …). These assets are aging and may present a risk of failing in groups at the same time. The company is not satisfied with its current replacement policy.

• Objective: develop a least-cost strategy for repair/replacement of these assets

• Specify a forecast of the expenses by category associated with this strategy

5

Aspects of the Aging Asset Problem• Optimal management of a single asset • Optimal policy for entire asset population• Cash flows for repair/replace for entire asset

population• Role of diagnostic tests

6

Two Fundamental Problems

• Optimal maintenance and replacement policy– Varies by asset class– Based on age, performance,

and condition information for individual assets

• Cash flow planning– Least cost replacement of

infrastructure inventory– Long term financial planning– Policy based on

maintenance and replacement policy for individual assets

Age

P(Fa

ilure

)

Cash Flow Forecast Versus Replace Policy

0.0

10.0

20.0

30.0

40.0

50.0

2000 2005 2010 2015 2020

Year

Ann

ual R

epla

ce C

ost

($00

0,00

0)

Replace at Age 40 Replace at Age 35 Replace at Age 30

7

Optimal Management of a Single Asset• Repair / replace strategy• Diagnostic tests• Individual performance—hazard function• State of asset

- Observable- Unobservable

• Solution: state-dependent optimal control

8

Diagnostic Tests• Several tests exist for each asset• Example: Underground Cable Tests

- Partial Discharge- Time Domain Reflectometry- Isothermal Relaxation Current- Dissipation Factor- Dielectric Spectroscopy- Wafer Test

9

Outline

• The Aging Asset problem• Substation transformer example—single asset• The need for a Learning Model – group exercise• Aging Asset model structure—single asset• Control of Asset Population - Cable Case Study• Organizational Issues

10

Substation Transformer Example

• Decisions– Repair (under what conditions?)– Replace– Maintain– Evaluative screening interval– Number of mobile, backup substations– Number of spares– Transfer load

11


• Economic Variables– Number of customers– Type of customers

• Residential• Commercial• Industrial/Critical

– Cost of Action• Maintain• Rebuild• Replace

12


• Observable States– Age– Peak load– Oil condition (result of chemical test)– Oil Temperature (result of temperature test)

• Observable states are “Decision Contingent”– Policies are contingent on observable states– Critical to model decision flexibility

13

Substation Transformer Example• Unobservable States reflect expert knowledge & problem structure

– Hot Spot TemperatureFunction of temperature and peak loadCalculable from current and load

– Oil condition (result of chemical test)– Oil Temperature (result of temperature test) – Degree of Global Embrittlement (the entire transformer)

Condition of insulationInfluenced by oil temp (temp can “half or double insulation life”)

– Degree of Local Embrittlement (around hot spot)Influenced by hot spot temperature

– Effective AgeInfluenced by local and global embrittlement

– Company-specific deviation from Industry Bathtub Curve

14

Relationship among observable and unobservable states

OilCondition

PeakLoad

BathtubCurve

EffectiveAge

Failure

Decision

OilTempHot Spot

Temp

GlobalEmbrtmntLocal

EmbrtmntAge

15

Outline

• The Aging Asset problem• Substation transformer example—single asset• The need for a Learning Model – group exercise• Aging Asset model structure—single asset• Control of Asset Population - Cable Case Study• Organizational Issues

16

Inference Example: Why we need Learning Model

• Uncertain Event: “Terrorist attack in USA in next week”

– You believe the probability is about 1%– Simulation: Toss coin ten times, observe zero or one

head

17


• Uncertain Event: “Terrorist attack in USA in next week”You believe the probability is about 1%

• Now you receive information: FBI/CIA Intelligence Report2 possible results: Threat “Credible evidence of attack ”

No Threat “ No credible evidence ”

18


• Uncertain Event: “Terrorist attack in CA in next week”You believe the probability is about 1%

• Now you receive information: FBI/CIA Intelligence Report2 possible results: Threat “Credible evidence of attack ”

No Threat “ No credible evidence ”

• Suppose you believe the agencies to be highly reliable: Given Attack 99% chance agencies pick up ThreatGiven No Attack 99% chance agencies say No Threat

19

Why we need Learning Model

Probability calculation:

• Given “Threat” Probability of Attack = ?

– (Or: How many tosses and how many heads?)

• Given “No Threat” Probability of Attack = ?

– (Or: How many tosses and how many heads?)

20

Why we need Learning Model

Simple probability calculation:

• Given “Threat” Probability of Attack = 0.50 (!!)

– One coin, one head

• Given “No Threat” Probability of Attack = .0001

– 17 coins, zero or one head

• Very strong empirical evidence that,“Humans are not very good at information processors”

21

Outline

• The Aging Asset problem• Substation transformer example—single asset• The need for a Learning Model – group exercise• Aging Asset model structure-single asset• Control of Asset Population - Cable Case Study• Organizational Issues

22

Aging Asset Model Structure-Single Asset

• Two model forms to fit problem characteristics• Planning Period

– Uncertainties and decisions may change year to year– Have detailed knowledge about near-term issues– One-of-a-kind events may be on the horizon– Stochastic dynamic program yields Time- and State-

dependent policy

• Post-Planning Period– Have less detailed year-to-year knowledge – Still need to capture long term policy– Markov decision process yields State-dependent policy

23

Model linkages

Near Term Long TermTime-varying Model Steady State Model

Age: 16

Condition: Poor

Last Temp: Hi

Age: 14

Condition: Good

Last Temp: Low

Each terminal node of the time-varying model plugs into a state

of the steady-state model

24

Outline

• The Aging Asset problem• Substation transformer example—single asset• The need for a Learning Model – group exercise• Aging Asset model structure-single asset• Control of Asset Population – Cable Case Study• Organizational Issues

25

Optimal Population Management –Underground Cable Case Study

• Compare optimal policy (replacement interval plus allowable failure history)

and• Non-replacement (continued repair)• Variation by type• Variation by capital cost, o&m, customer values

26

Population Management Model Logic

Read InputParameters

Create CableInventory Matrix Set Year to 1 Age Inventory

Replace CableBased on Retire

Policy & Fail Policy

Determine CableFailures

Calculate CapitalReplace Costs

Calculate O&MRepair Costs

Calculate CustomerOutage Costs

Increment ForecastYear

If Done

No

Report Results Yes

27

Population Hazard Function -- Example

• A1: Failures:2AWG 1/0 2/0 4/0 500 750

394 152 61 24 * 28• A2: 85% of failures occur in cable installed prior to

1983 (non-tree retardant)• A3: 16% of failed segments experience additional

failure in the same year* 500 assumed to fail as 750

28

Hazard Function (cont’d)

• A4: Two functional forms

hss

T

hss (1+α(t-T))post -1983

T-δ

hss

hss (1+ 2α(t-(T-δ)))

pre-1984

t

t

29

Sample Results Summary – Optimal Management of Cable Inventory

.0150+ Years, 3 Fails

750 New

.2930 Years, 3 Fails750 Old

.0250+ Years, 3 Fails

#2 New

1.040 Years, 2 Fails#2 Old

PV CostsOptimal PolicyCable Type

30

Sample Results - #2 Cable Old

Cable #2- Old

0.0

5,000.0

10,000.0

15,000.0

20,000.0

25,000.0

30,000.0

35,000.0

40,000.0

45,000.0

50,000.0

20 25 30 35 40 45 50 55 60

Policy - Replace Age

Num

ber

Miles Replaced Segment Failures

Cable #2- Old

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

20 30 40 50 60

Replace Age

PV C

osts

Replace Costs ($000) Repair Costs($000)

Customer Costs($000) Total Costs($000)

31

Sample Results – #2 Cable New

Cabke #2 New

0.0

200.0

400.0

600.0

800.0

1,000.0

1,200.0

20 25 30 35 40 45 50 55 60


Num

ber


Cable #2- New

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

20 30 40 50 60

Replace Age

PV C

osts

Replace Costs ($000) Utility Failure Costs ($000)

Customer Failure Costs ($000) Total Costs($000)

32

Sample Results - 750 Cable Old

Cable 750 Old

0.0

100.0

200.0

300.0

400.0

500.0

600.0

700.0

800.0

900.0

20 25 30 35 40 45 50 55 60


Num

ber


Cable 750 Old

0.0

0.2

0.4

0.6

0.8

1.0

1.2

20 30 40 50 60

Replace Age

PV C

osts



33

Sample Results - 750 Cable New

Cable 750 New

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

180.0

200.0

20 25 30 35 40 45 50 55 60


Num

ber


Cable 750 New

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

20 30 40 50 60

Replace Age

PV C

osts



34

Sample Results - Total Costs - Continued Repair & Optimal Replace – #2 Old

Never Replace Versus Optimal Policy - #2 Old

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022

Year

Tota

l Cos

ts

Never Replace Optimal Policy

35

Sample Results - Total Costs - Continued Repair & Optimal Replace – #2 New

Never Replace Versus Optimal Policy - #2 New

0.50

0.70

0.90

1.10

1.30

1.50

1.70

1.90

2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022

Year

Tota

l Cos

ts


36

Sample Results - Total Costs - Continued Repair & Optimal Replace – 750 Old

Never Replace Versus Optimal Policy - 750 Old

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022

Year

Tota

l Cos

ts (0

00)


37

Sample Results - Total Costs - Continued Repair & Optimal Replace – 750 New

Never Replace Versus Optimal Policy - 750 New

0.95

0.97

0.99

1.01

1.03

1.05

1.07

1.09

2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022

Year

Tota

l Cos

ts


38

Outline

• The Aging Asset problem• Substation transformer example—single asset• The need for a Learning Model – group exercise• Aging Asset model structure-single asset• Control of Asset Population – Underground Cable Case

Study• Organizational Issues

39

Organizational Issues—What is aging asset management?

• What is the objective?• Process or analysis?• Analytic requirements

– Data issues• Failure rates• Equipment characteristics• Customer needs and values• Utility costs• Constraints (budgets, operating, …?)

– Commitment to analysis

1

Aging Assets Case StudyPSEG Air Breakers

Steve Chapel

Charles D. Feinstein

May 2004

2

Background

♦ Air Breakers: – installed in the PSEG 138—230—500 kV transmission system

– 25 or 26 air breakers installed in the part of the system under study

– breakers are all approximately 30 years old. None has failed yet.

♦ Two failure modes

– catastrophic mode: pressurized porcelain tube blow up and shatters, scattering parts up to 700 feet away

– non-catastrophic mode: the breaker can develop air leaks (in either 500psi or 2000psi systems) which will induce a compressor failure and causes and interruption. The cause of this failure mode is worn parts, such as aging gaskets. The critical time for this failure mode is dependent on asset age, and appears to be in the range 8-10 years old.

3

Background cont’d♦ The impact of a failure on customers varies

♦ The current maintenance cycle

– yearly inspection and maintenance

– special procedures every three, four, six, and nine years

– field rebuild every nine years

4

Decision Problem

♦ The problem: PSEG does not have a repair / replace policy for air breakers– when to replace air breakers, – when to rebuild them – when to maintain – how best to treat air breakers that exhibit poor performance

♦Objectives: – Develop repair replace policy old air breakers that minimizes life

cycle costs including customer and utility costs– Evaluate the current policy of rebuilding air breakers on a nine-year

cycle

5

Decision Problem - cont.

♦ Decision - what to do with an old breaker?

– maintain the breaker without rebuilding it

– rebuild the breaker in place, which costs approximately $150,000

– refurbish the breaker in a shop

– purchase a new breaker, which costs approximately $300,000.

♦ Uncertainties (and relevant parameters describing the uncertainties):

– Probability of failure of new and old breakers as function of age

– Probability of failure as a function of time since rebuild

– Probability of failure conditional on a breaker performance - good or problem

6

Analysis - Approach

♦A multi-stage decision tree was used to evaluate and identify the least-cost repair / replace strategy.

♦Under this approach the user specifies all decisions, uncertainties and outcomes

♦ The method identifies the set of decisions over time that meet the objective

7

Analysis - Inputs

♦ Types of technology – Old breaker – New breaker

♦ Policy decisions: maintain, rebuild, replace with new

♦ Technology performance states:– Good– Problem– Failure– C-failure

♦ Decision stages: 8 stages, each stage is 3 years in length– Make a decision on a breaker then revisit every three years– Model behavior for 24 years

8

Inputs con’d - Probability Tables Old Tech. Performance

Technology Performance Probabilities (Old Breakers, Prior Performance = Good)

Technology Performance Probabilities(Old Breakers, Prior Performance = Problem)

Perf. State 1-3 4-6 7-9 10-12 13-15 16-18Good 0.8 0.75 0.6 0.4 0.4 0.4Problem 0.15 0.15 0.2 0.3 0.3 0.3Failure 0.05 0.05 0.15 0.2 0.2 0.2C-failure 0 0.05 0.05 0.1 0.1 0.1

Breaker Age (Decision Stage)

Perf. State 1-3 4-6 7-9 10-12 13-15 16-18Good NA 0 0 0 0 0Problem NA 0.7 0.65 0.5 0.5 0.5Failure NA 0.25 0.3 0.4 0.4 0.4C-failure NA 0.05 0.05 0.1 0.1 0.1

Breaker Age (Decision Stage)

9

Inputs con’d - Probability Tables New Tech. Performance

Technology Performance Probabilities (New Breakers, Prior Performance = Good)

Technology Performance Probabilities(New Breakers, Prior Performance = Problem)

Perf. State 1-3 4-6 7-9 10-12 13-15 16-18Good 0.95 0.99 0.99 0.99 0.99 0.95Problem 0.05 0.01 0.01 0.01 0.01 0.05Failure 0 0 0 0 0 0C-failure 0 0 0 0 0 0

Perf. State 1-3 4-6 7-9 10-12 13-15 16-18Good NA 0 0 0 0 0Problem NA 0.95 0.95 0.95 0.9 0.9Failure NA 0.05 0.05 0.05 0.1 0.1C-failure NA 0 0 0 0 0

10

Inputs cont’d - CostsTechnology Decision Costs

Technology Performance Costs

Decision Old Breakers New BreakersMaintain 50 2Rebuild 150 + O&M NAReplace 300 + O&M 300 + O&M

Technology

Performance Old Breakers New BreakersGood O&M O&MProblem 1 - 5 1 - 5Failure 10 - 50 2 - 50C-Failure 300 - 500 300

Technology

1

Strategic Reliability Analysis

Charles D. Feinstein & Peter A. MorrisVMN Group LLC

AndSteve Chapel

S.Chapel Associates

May 2004

2

Outline

Analytical Modeling of ReliabilityStrategic FocusBeyond Distributions on Averages

The Importance of Rare but Catastrophic EventsPitfalls of Historical Analysis and the need for Scenario ThinkingThe Strategic Reliability Problem

3

Analytic Reliability Modeling

Classic reliability models don’t address the full range of customer outcomes

Analytic models based on classic reliability theory estimate reliability indices, which measure system averages Simulation models simulate variations in system averages

Simulations of average measures like SAIFI greatly underestimate the range of outcomes for customers Simulation’s classic problems: “Simulation is not enumerative and may overlook rare and important events”

4

Analytic Reliability Modeling (cont’d)

Current models are not strategically focused They do not focus on the rare but catastrophic events of most concern to utilities and the publicThey do not represent the true range of uncertaintyThey do not optimize strategy

5

Analytic Reliability Modeling (cont’d)

Distributions on system averages greatly understate the range of outcomes

Mean outage duration = 4, σ = 5SAIDI for 100 customers: µ = 4, σ = 0.5

-0.10

0.10.20.30.40.50.60.70.80.9

1

0 2 4 6 8

Hours per Outage

Distribution of MeanOutage DurationDistribution of OutageDuration

6

The Importance of Rare but Catastrophic Events

Probabilities of rare events can drive strategyInventory problems are an example

Costs are relatively flat for range of policies under normal operationsOptimal policy based on costs and frequency of stockout or service interruption, a rare event

7

The Importance of Rare but Catastrophic Events (cont’d)

Policy consequences of probabilities of rare events—inventory exampleOptimal policy can change greatly as frequency changesIndicates what is important to know (what is true frequency?)

OptimalInventoryLevel

Interruption Rate (Time between interruptions)

1 2 3 4 5 6 7 8 9 10 11 12 13

8

Pitfalls of Historical Analysis

Statistical analysis alone is inappropriate for predicting rare, one-of-a-kind eventsModels must be augmented with expert judgmentProcedures exist for encoding expert judgment into usable model inputs

9

Recent events highlight the importance of scenario thinking

1900 1920 1930 1940 1960 1980 2000

History of Terrorism in the US

Victims ofTerrorism

1000

2000

3000

4000

5000

10

The Strategic Reliability Problem

Catastrophic events result from two fundamental causes

Added stress on subsystem components associated with failure of a subset of components in the the same subsystemLinkages among subsystems permit failure consequences to cascade—complex systems can fail in unanticipated ways

11

The Strategic Reliability Problem (cont’d)

Data is insufficient to characterize catastrophic event ratesScenario planning is requiredDetailed models of subsystem behavior do not address the strategic questionsPolicy failure is a failure of the imagination

12


Fundamental Modeling IssuesCommon Mode FailuresCascading Failures (the avalanche)Complexity and Dependency

13

Common Mode Failures

MULTI-CIRCUIT PROBABILISTIC RELIABILITY ANALYSIS Two Circuits, each described by a Markov Process

Circuit A up down Circuit B up down

up 0.990 0.010 up 0.980 0.020

down 0.600 0.400 down 0.300 0.700

Specify Special Transition Definition Specify Common-Cause Failure Probability

From To 0.0

Model ResultsPlan Period 60 Years

Steady State Annual Frequency of Catastrophes 0.001Number Probability

0 0.9568718Steady State Average Years Between Catastrophes 976 Years 1 0.0302407

2 0.00904793 0.00269894 0.00080275 0.00023816 0.00007047 0.00002088 0.00000619 0.0000018

10 0.0000005

WHEN THE CIRCUITS ARE INDEPENDENT, THE TIME BETWEEN FAILURES IS ALMOST 1000 YEARS

Probability Distribution

A down B downany state

14

Common Mode Failures (cont’d)

MULTI-CIRCUIT PROBABILISTIC RELIABILITY ANALYSIS Two Circuits, each described by a Markov Process

Circuit A up down Circuit B up down

up 0.990 0.010 up 0.980 0.020

down 0.600 0.400 down 0.300 0.700

Specify Special Transition Definition Specify Common-Cause Failure Probability

From To 0.3

Model ResultsPlan Period 60 Years

Steady State Annual Frequency of Catastrophes 0.035Number Probability

0 0.3745Steady State Average Years Between Catastrophes 29 Years 1 0.1675

2 0.13083 0.09844 0.07185 0.05116 0.03567 0.02438 0.01639 0.0108

10 0.0070

WHEN THE CIRCUITS ARE NOT INDEPENDENT, THE TIME BETWEEN FAILURES IS ENORMOUSLY REDUCED

COMMON-CAUSE POSSIBILITIES ARE FAR MORE CRITICAL TO SYSTEM FAILURE THAN THE RELIABILITY OF THE SYSTEM COMPONENTS

Probability Distribution

A down B downany state

15

Cascading Failures (the avalanche)

Avalanche Model -- An Example of Independent Failures, No Interaction

Number of Linked Assets 10 degree of system complexity (number of connected rocks)

Asset 1 Asset 2 Asset 3 Asset 4 Asset 5 Asset 6 Asset 7 Asset 8 Asset 9 Asset 10Age 20 20 20 20 20 20 20 20 20 20

Independent Failure Rate 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050incidence rates (size of "rocks")

201 Failure 2 Failures 3 Failures 4 Failures 5 Failures 6 Failures 7 Failures 8 Failures 9 Failures 10 Failures

Trigger Rate for Future Events 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000degree of dependency (strength of avalanche)

0.1001.0

24

Catastrophe = 3 or more Failures disaster management flexibility (avalanche warning system)

SystemFailures Probability

0 0.60781 0.25052 0.09513 0.03284 0.01025 0.00286 0.00067 0.00018 0.00009 0.000010 0.0000

Total 1.0000

Catastrophe Rate 0.0465

RESULTS

SYSTEM RISK CHART

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 1 2 3 4 5 6 7 8 9 10

NUMBER OF FAILED ASSETS

PRO

BAB

ILIT

Y

16

Cascading Failures (the avalanche) (cont’d)

Avalanche Model -- An Example of Interrelated Failures

Number of Linked Assets 10degree of system complexity (number of connected rocks)

Asset 1 Asset 2 Asset 3 Asset 4 Asset 5 Asset 6 Asset 7 Asset 8 Asset 9 Asset 10Age 20 20 20 20 20 20 20 20 20 20

Independent Failure Rate 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050 0.050incidence rates (size of "rocks")

201 Failure 2 Failures 3 Failures 4 Failures 5 Failures 6 Failures 7 Failures 8 Failures 9 Failures 10 Failures

Trigger Rate for Future Events 0.100 0.200 0.400 0.800 1.000 1.000 1.000 1.000 1.000 0.000degree of dependency (strength of avalanche)

0.1001.0

24

Catastrophe = 3 or more Failures disaster management flexibility (avalanche warning system)

SystemFailures Probability

0 0.60781 0.09712 0.03323 0.00524 0.00005 0.00006 0.00007 0.00008 0.00009 0.000010 0.2567

Total 1.0000

Catastrophe Rate 0.2619

RESULTS

SYSTEM RISK CHART

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 1 2 3 4 5 6 7 8 9 10

NUMBER OF FAILED ASSETS

PRO

BA

BIL

ITY

17

Complexity and Dependency

Strategic Design Issue: Complexity versus Dependency

Number of Independent Assets ClustersNumber of Links in ClustersStrength of Links Within Clusters

Example: 10 Mildly Linked Assets,

10 Clusters 5 Clusters 2 Clusters 1 Cluster0 Failures 0.608 0.608 0.608 0.608

1 Failure 0.251 0.117 0.109 0.097

2 Failures 0.095 0.210 0.039 0.033

3 Failures 0.033 0.031 0.008 0.005

4 Failures 0.010 0.028 0.001 0.000

5 Failures 0.003 0.003 0.196 0.000

6 Failures 0.001 0.002 0.018 0.000

7 Failures 0.000 0.000 0.005 0.000

8 Failures 0.000 0.000 0.001 0.000

9 Failures 0.000 0.000 0.000 0.000

10 Failures 0.000 0.000 0.016 0.257

Total 1.000 1.000 1.000 1.000

Catastrophe 0.04654 0.06493 0.24433 0.26188

18

The Strategic Reliability Problem (cont’d):Current state of the art is out of balance

Great emphasis on solving tactical problems, supported by lots of data and detailed modelsLittle emphasis on solving strategic problems, very little data, no accepted modeling approaches (that we are aware of), so we are developing:

Strategic ParadigmsAnalytic ModelsKey Data and Judgments

Dominant costs and customer effects are on the strategic side (latest estimate of cost of NE blackout is in the neighborhood of $10 billion!!)Tactical Effort > Strategic Effort, but

Tactical Importance < (<?)Strategic Importance

19


Simplified versions of some of our models are available in a software demonstration


SUMMARY




May 2004

2

Summary

What is the T&D asset management problem?What are the specific asset management decisions that companies face?What are the useful analytic tools?What data is required to apply tools and make decisions?What are the organization issues?

3

What Is The T&D Asset Management Problem?

Specific T&D Aspects?Maintaining the existing infrastructureExpanding system to meet future needsMeeting customer needs Minimizing costsMaking money

Analytical Aspects?Defining objectivesMeasuring valueAssessing uncertainty and riskChoosing and appropriate time horizonForecasting

4

What Are The Specific Asset Management Decisions?

Repair / Replace?Prioritization?Capacity?Other

5

What Are The Useful Analytic Tools?

For repair / replace?For prioritization?For capacity decisions?For setting budgets?For forecasting cash flows?

6

What Data Is Required To Apply Tools And Make Decisions?

For repair / replace?For prioritization?For capacity decisions?For setting budgets?For forecasting cash flows?

7

What Are The Organization Issues?

Link between specific decisions and corporate objectives / strategyCorporate appreciation of specific customer and engineering issues Incorporating multiple stakeholders valuesRole of process and role of analysis

8

Some advice for those who use analytic tools

Use a tool with an “enterprise-level” view of asset management.Choose a tool capable of capturing all critical considerations. Commonly ignored considerations include:

soft benefits investment urgency (as opposed to investment value) risk (especially “correlated risks”)sequencing and other project interdependencies

9

Some advice for those who use analytic tools (continued)

Don’t use tools that have no basis in accepted mathematical theories. If in doubt, get an independent review.Plan on the need for education and training. Having a thorough understanding of concepts is as important as knowing how to use the tool.Use a phased approach to implementation. Start with one department, conduct lessons-learned review, and make changes before you move to the next level.

10

Some advice for those who use analytic tools (continued)

Make sure that all stakeholders have necessary “buy in” and confidence—otherwise, insufficient effort will be devoted to generating inputs and/or model recommendations won’t change decisions.If the answer seems wrong, don’t trust it (and get a new model!). A tool is an aid (not a substitute) for decision making.

11

Revisiting Asset Management Beliefs – Fact or Fallacy?

Problems can be solved by organizational change and asset management teams"Cost Benefit analysis, the "tried and true" method for ranking projects is perfectly reasonable way to select projects to fund."The first important step is to gather dataAll projects can be valued using the same aggregate measures (i.e. $)Projects are risky because of uncertain financial consequences Beta is an appropriate way to measure project risk

12

Fact or Fallacy cont.

An important objective is near-term profitability The biggest problem for asset management is insufficient data. ROI is the good metric for evaluating investments. Ranking is a good way to prioritize projects. Strategic alignment is a good metric for prioritizing projects. Balanced scorecards are a good way to prioritize projects. Hurdle rates are a good way to account for risks.


SOME FINAL THOUGHTS




May 2004

2

What Next?

What are some of the things you might think about doing?Problem assessment: do you have any of these problems?Time assessment: if you own a problem, when do you need a solution?Method assessment: if you need a solution, do you have a method for finding one?

3

We believe…

This EUCI Seminar is not a sales pitchWe attempted to teach and explain principles and methods These are challenging problems both to formulate and to solveMost utilities have a difficult time solving these problems by themselvesIf you have the problem, we can help you solve it

4

What Are Possible Ways to Proceed?

In-house seminarsProblem assessmentMethods and practice assessmentSoftware demosTest CasesWorkshopsCoachingConsulting Projects

5

And In Closing

Documents

Analytical Tools for Asset Management