22
Risk Modelling of a Notional Mega Project Cost Estimate 11 Nov 2011

Palisade2011_PaulGill

Embed Size (px)

DESCRIPTION

hh

Citation preview

  • Risk Modelling of a Notional Mega Project Cost Estimate 11 Nov 2011

  • *

    Overview

    Mega Projects by their very nature consist of smaller individual projects that integrate together to form a whole.

    In a similar fashion the estimate for such a project also comes from a multitude of diverse disciplines with differing sources and work practices.

    This paper will use a case study to show how a risk model for the Cost Estimate was built in Palisade @Risk to identify appropriate levels of Contingency and Management Reserve.

    Inputs to the model were taken from a variety of sources including

    Control EstimateSystemic risk modelsQuantitative schedule risk modelsRisk ranging workshops Various risk registers
  • *

    Contingency & Management Reserve

    Contingency

    Contingency is the amount of money used in the estimate to deal with the uncertainties inherent in the estimating process. Contingency is required because estimating is not an exact science.

    The amount of infill and concrete to cover an underground pipe, the number of man-hours to complete the task, and the actual all in labour rate, are all best estimates until the work is complete

    Management Reserve

    An amount added to the estimate to allow for specific risks that may or may not occur that are within the projects control or influence. Risk is defined as an undesirable potential outcome and its probability of occurrence

    Does not include force majeure, currency risk, political risk etc

    Scope

    Both amounts are based on a defined scope. Should scope change , the estimate must be revised to reflect such changes in scope.

    Neither Contingency nor Management Reserve are a source of funds to cover scope changes

  • *

    Mega Project vs Project Portfolio

    Key Differences

    So why is a Mega Project so different to a Portfolio of Projects being executed by a Company.

    Key differences are:

    Same key personnelMega project is a number of inter related projects with numerous key interfaces. Any delay or changes to one can have a significant effect on one or the othersSite Wide Services common to all sub projectsSame geographical location

    Effect

    Must not underplay the significance of these differences. Sub projects must not be modeled in isolation

    Liberal use of correlation between sub projects to avoid nodal biasSchedule must be modeled at the mega project level ensuring all interfaces are included.
  • *

    Identifying the Risks

    The Risk Register

    Prior to estimating contingency or otherwise quantifying risk impacts, the risks must first be identified and then logged in the Risk register. Husky uses a standard 5 x 5 risk Matrix as a Probability Impact Diagram for Project Risk. Impacts are defined specific to project

  • *

    Typical Risk Register

    *

  • *

    Typical Boston Square

  • The Cost Model

    *

    Main input to the Cost Model was the Control Estimate. However elements of the estimate were modelled using a variety of techniques and then brought into an overall Cost Risk model

    Risk ranging workshops using range estimating techniquesSystemic risk modellingQuantitative schedule risk modelsSpecific risk modelling
  • Traditional Probability Distribution Functions

    @Risk contains a wealth of distribution functions.

    Most are useful to the in depth simulation requiring sophisticated tools.

    Only a few are suitable for general cost modelling

    *

  • *

    Typical Probability Distribution Functions

    Risk Uniform

    Used when we have no idea what the value is between two limits


    Risk Triangle

    Most Popular distribution to show Most Likely value tapering to a Min and Max


    Risk Trigen

    Modification of Triangle

    Allows for a finite probability of achieving Min & Max Values


    Fundamental Flaw of Triangle and Trigen is when the distribution is skewed

  • Most Likely =5, Min =4, Max =15

    Most Likely = 5

    Mean = 8

    P50= 7.58

    *

    Most Likely = 5

    Mean = 6.5

    P50= 6.16

  • *

    AACE International Recommended Practice No. 41R-08

    RISK ANALYSIS AND CONTINGENCY DETERMINATION USING RANGE ESTIMATING

    Monte Carlo software for risk analysis requires identification of a probability density function (PDF) for each critical item. In rare instances the behavior of a critical item is known to conform to a specific type of PDF such as a lognormal or beta distribution, which reflects items that may skew heavily to one side of a distribution. However in most instances it is unlikely that the actual type of PDF that truly represents the item is known. Thus a reasonable approximation is to use either

    - Triangular Distribution

    - Double Triangular Distribution

    In most cases, the double triangular distribution is a better approximation since it can be made to conform to the implicit skew of the project teams probability assessment. The double triangle allows the risk analyst to use the probabilities which the project team believes are reasonable rather than letting the triangular distribution dictate a probability which, more often than not, is invalid.

  • Double Triangle Method

    *

    a

    c

    b

    Area = Underrun Probability

    Area = Overrun Probability

    Probability Density

    Random Variable x

    Most Likely = 5

    Mean = 6.5

    P50= 5.0

  • Double Triangle Method

    F1= 2*Urun/Min

    F2= 2*(1-Urun)/Max

    RF=1+RiskGeneral(Min,Max,{0,0},F1:F2,RiskStatic(0))

    *

    Min

    ML

    Max

    Urun

    Probability Density

    F1

    F2

    1-Urun

  • *

    Schedule Risk

    Quantitative Schedule Risk Analysis

    Husky uses Primavera P6 to schedule all projects greater than $5MM. Oracle Primavera Risk, formerly known as Pertmaster, is used for quantitative schedule risk analysis (QSRA)

    To bring in the cost element of schedule delay, the results of the QSRA were brought into the cost model as a set of percentiles days delay from P0 to P100.

    Use of the @Risk Fit Manager was used to fit the best curve to this profile, with options set to update the curve each simulation. This simplified the update process each time the schedule model changed.

    The schedule @Risk function was then multiplied by another @Risk function that represented the uncertainty in cost per day to give an effective cost for schedule slippage


  • Schedule Cost Risk

    *

    =RiskLoglogistic(-1063,1088.2,38.361, RiskFit("Schedule","RMSErr"), RiskName("Schedule"), RiskStatic(0))

    27/06/20130%25-Feb-13-1225%07-May-13-5110%23-May-13-3515%04-Jun-13-2320%13-Jun-13-1425%21-Jun-13-630%28-Jun-13135%05-Jul-13840%11-Jul-131445%17-Jul-132050%22-Jul-132555%28-Jul-133160%02-Aug-133665%09-Aug-134370%16-Aug-135075%23-Aug-135780%01-Sep-136685%11-Sep-137690%24-Sep-138995%12-Oct-13107100%24-Dec-13180
  • Use of a Systemic Risk tool

    *

    Systemic Risk

    Systemic risk drivers such as the level of project scope definition affect individual, disaggregated estimate line items in a way that is hard to see and predict.

    Best practice is to address systemic risk drivers using empirical knowledge (from historical data) to produce stochastic models that link known risk drivers (e.g. scope definition) to bottom line project cost growth. It was decided that this approach would be best suited for the Process Plant, which formed the major part of the cost estimate.Conquest Consulting Group , who have wide experience in the use of such tools, were engaged to construct a parametric model which used a series of questionnaires to the project team based on :

    - Contractor Organisation

    - Contractor Experience

    - Project Planning

    - Execution Strategy

    - Scope Definition

    Fit Manager was again used to integrate the results from the parametric model into the overall model
  • Systemic Risk Questionnaire

    *

  • *

    Specific Risk Input

    Probability and Impact

    Specific Risks consist of both a probability of occurrence, and an impact should they occur

    Risk Probability modeled as simple binomial to simulate Yes/NoRisk Impact modeled as a PDF to give uncertainty of the impactRisk outcome modeled as a product of Probability (0 or 1) and Impact

    Problems

    Prior to release of @Risk 5.0 there were inherent problems using this method in that Tornado Diagram showed both Probability and Impact as two separate risk inputs

    RiskMakeInput

    @Risk5.0 onwards allowed for use of RiskMakeInput to combine the two together.

    Use extract from Risk Register to directly map risks into the model

    Risk Amount = RiskMakeInput( (RiskBinomial(1, Prob,RiskStatic(0))*

    RiskUniform( Min,Max)),RiskName(Description))


  • *

    Risk Results

    Summary Results

    Individual Risk models were established on separate MSExcel sheets and outputs summarised on tabular results sheet

    Extensive use made of the @Risk function RiskPercentile to provide tabular output that could be copied and pasted direct into reports

    Note the values in the above table are for a fictitious project

  • *

    Summary

    The cost risk model built in @Risk used a variety of techniques to both represent uncertainty from ranging workshops and from other studies. Highlights include:

    Use of the RiskGeneral function to mimic the Double Triangular Distribution recommended by AACE for range estimatingUse of the Fit Manager to replicate output from other models as input to the overall risk model, in particular:
    - Results from a quantitative schedule risk analysis
    - Results from systemic parametric risk analysis

    - Automatic recalculation of the curve fit on change of input data

    Use of RiskPercentile and other output functions to provide templated report formats that can be copied and pasted direct into presentational materials


  • *

    Any Questions

    Likelihood of Occurrence

    Consequence

    Very Low

    Low

    Medium

    High

    Very High

    Very High

    5A

    5B

    5C

    5D

    5

    E

    High

    4A

    4B

    4C

    4D

    4E

    Medium

    3A

    3B

    3C

    3D

    3E

    Low

    2A

    2B

    2C

    2D

    2E

    Very Low

    1A

    1

    B

    1C

    1D

    1E

    Priority Action Setting

    Critical

    Immediate Action must be taken to Prevent or Mitigate the

    Risk

    Serious

    Mitigation Action

    Required

    Moderate

    Mitigation Strategies to be investigated

    Acceptable

    Be aware of Potential Mitigation Strategies

    MinMaxUrunF1F2RF

    -10%20%30%671.00

    Case Description:

    Date:

    2

    Enter Contractor Bid Value

    1,079,000

    ($ thousands)Currency

    Canadian$

    Enter Execution Schedule Bid Duration

    39.8

    (months)

    3

    Rating

    Overall Contractor Capabilities

    3

    Contractor Organizational Structure

    3

    Alignment of JV Partners

    3

    Key Project Personnel Proposed

    4

    Contractor Ramp-up Strategy and Capability

    3

    Contractor Resource Capability and Availability

    4

    Alignment with Subcontractors and Suppliers

    3

    Contractor Organization Evaluation

    3.3

    Ratings: 0=N/A, 1=Greatly Exceeds, 2 = Exceeds, 3 = Meets, 4 = Does Not Meet, 5 = Fails (EXPECTATIONS)

    added note:

    Does not meet expectations with regard to the key personnel proposed for the project (years of experience, applicable experience, guarantee of

    staying with the project, etc.)

    added note:

    Meets expectations with regard to the proposed ramp-up strategy and capability to adequately resource the project in a timely fashion.

    added note:

    added note:

    added note:

    added note:

    Meets expectations with regard to the proposed organization structure to support the project.

    BID MATURITY WORKSHEET

    SNAM

    9/30/2010 & 10/22/2010

    Meets expectations with regard to overall capabilities (including financial strength, reputation, experience, resources, etc.).

    Contractor Organization Evaluation

    Meets expectations with regard to alignment of the contractor with sub-contractors and suppliers.

    added note:

    Meets expectations with regard to clear and documented alignment of JV Partners.

    Does not meet expectations with regard to resource capabilities and availability (such as number of in-house resources and capability to contract

    resources).

    RiskIDDescriptionScoreProbMin $Max $Risk amount $

    FFH-013Construction in advance of sufficient engineering

    A3

    5%2,000,0003,500,000

    0

    $MM

    Estimate

    Risked Value

    50%

    70%

    90%

    P50

    P70

    P90

    Process Plant

    $918.3

    $918.3

    $1,038.2

    $1,107.7

    $1,209.4

    13.1%

    20.6%

    31.7%

    SubSurface Development

    $1.6

    $1.6

    $1.8

    $2.0

    $2.3

    16.4%

    27.2%

    42.9%

    Drilling & Completions

    $193.8

    $193.8

    $208.4

    $217.0

    $229.3

    7.6%

    12.0%

    18.3%

    Infrastructure

    $124.4

    $124.4

    $133.4

    $137.4

    $143.3

    7.2%

    10.4%

    15.2%

    Midstream

    $30.2

    $30.2

    $33.2

    $34.6

    $36.6

    9.9%

    14.6%

    21.2%

    Business & System Management

    $11.3

    $11.3

    $12.8

    $13.2

    $13.9

    13.2%

    16.8%

    22.7%

    Owners Costs

    $156.6

    $156.6

    $171.7

    $176.4

    $183.2

    9.6%

    12.6%

    17.0%

    Additional Specific Risk

    $.0

    $.0

    $22.3

    $28.2

    $38.1

    TOTAL PROGRAM

    $1,436.2

    $1,436.2

    $1,623.6

    $1,693.4

    $1,798.6

    13.0%

    17.9%

    25.2%

    Management Reserve

    $.0

    $.0

    $17.9

    $30.7

    $59.6

    Overall Total

    $1,436.2

    $1,436.2

    $1,648.2

    $1,717.8

    $1,825.1

    14.8%

    19.6%

    27.1%