National Academies Press: OpenBook
« Previous: 6 RISK ASSESSMENT
Page 69
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 69
Page 70
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 70
Page 71
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 71
Page 72
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 72
Page 73
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 73
Page 74
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 74
Page 75
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 75
Page 76
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 76
Page 77
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 77
Page 78
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 78
Page 79
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 79
Page 80
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 80
Page 81
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 81
Page 82
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 82
Page 83
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 83
Page 84
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 84
Page 85
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 85
Page 86
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 86
Page 87
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 87
Page 88
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 88
Page 89
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 89
Page 90
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 90
Page 91
Suggested Citation:"7 RISK ANALYSIS." National Academies of Sciences, Engineering, and Medicine. 2012. Guide for the Process of Managing Risk on Rapid Renewal Projects. Washington, DC: The National Academies Press. doi: 10.17226/22665.
×
Page 91

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

67 INTRODUCTION The tasks of identifying, assessing, and managing risk for rapid renewal projects can produce results useful to project risk managers, in helping to understand and optimize project performance. However, there is another very valuable process within the sphere of risk management, risk analysis, which can provide additional valuable information to project managers when projects are more complex or the information required for decisions must be more precise. Objectives Risk analysis starts with the results from structuring, risk identification, and risk assessment, as described in the previous chapters. Risk analysis then expands on those elements and combines them to quantify the key project performance measures, such as project cost and sched- ule, considering risk as well as base. This can be done in terms of mean values (as discussed in Chapter 6) or more completely in terms of full uncertainty (e.g., Figure 7.1). Results from risk analysis can then be used to help make important project decisions because they contain more detail and information than do risk assessments. Hence, the primary objectives for risk analysis are to • Adequately quantify uncertainty in the project performance measures, such as project inflated year-of-construction cost and completion date, appropriately con- sidering risks as well as the base uncertainties; 7 RISK ANALYSIS Adequately but efficiently (a) quantify uncer- tainties in (and correlations among) inputs (including risks and opportunities); (b) prop- agate those uncertainties through to out- puts (e.g., project cost and schedule); and (c) quantify sensitivity.

68 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS • Adequately (a) quantify the likelihood for achieving existing budgets and mile- stones or (b) establish budgets and milestones (including contingencies) for a desired reliability or confidence level (e.g., 80% chance for success); and • Adequately quantify the sensitivity of those project performance measures to the individual risks and base uncertainties, which provides additional information for risk management planning. Ideally, this would be done not only from a current perspective but also projected to various milestones to determine remaining costs and schedule to finish (e.g., to establish defensible contingency drawdown requirements). Another goal is to complete this step efficiently, producing defensible as well as accu- rate results that are compatible with the other steps of the process. How this information will be used will determine the requirements and the level of effort (which can be signifi- cant) for this step. However, adequate quantification of the significant uncertainties in the various base and risk factors and development of an appropriate risk model that can be easily updated are keys to successfully completing this step. Philosophy and Concepts Performance measures can generally be adequately estimated as a function of specific factors. For example, total project cost is simply the sum of all of the various costs, both base and realized risks. As another example, the project completion date can be deter- mined by critical path analysis, based on activity durations (both base and realized risks) and precedence requirements (including lags and external milestone dates). Typically, however, there is significant uncertainty in what those factors will be (especially risks, which might or might not occur), which in turn results in significant uncertainty in what the performance measures will be. Generally (as discussed in Chapter 6), mean values of the performance measures can be adequately approximated as a function of the mean values of those various factors. However, the determination of the full uncertainty in performance measures requires more sophisticated analysis, which can be done in vari- ous ways with different levels of accuracy and defensibility, and thus effort. The types of results produced by risk analysis are illustrated later in this chapter by example. The various important concepts associated with risk analysis include • Qualitative versus quantitative assessment; • Uncertainty description; • Performance measures; • Deterministic versus probabilistic analysis; • Risk-based versus non–risk-based analysis; • Time-variable versus time-independent analysis; • Decoupled versus integrated analysis; • Initial versus updated analysis; and • Levels of detail, accuracy, defensibility, and effort.

69 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Qualitative Versus Quantitative Assessment This was addressed in Chapter 6 with respect to risk assessment. For risk analysis as described in this chapter, quantitative assessment is required, generally including ex- plicit quantification of significant uncertainties (in terms of probability distributions) and correlations for input variables. The discussion in Chapter 6 focused on mean- value assessments, which are appropriate for some applications but ignore uncertain- ties and correlations. Uncertainty Description Uncertainties can be described in terms of “probability distributions,” which express the relative likelihood of any one particular value for a factor that has a set of possible values. The uncertainty in the value of a particular factor can be expressed in different ways, depending on the nature of that factor (Figure 7.1): • Two possible values (e.g., yes or no)—probability (Figure 7.1a); • Discrete set of possible values (e.g., several ranges of values, or scenarios)— discrete distribution (Figure 7.1b), which in turn can be combined into two states (e.g., either more or less than a particular value, or either one or the other subset of scenarios); and • Infinite set of possible values (e.g., cost)—continuous distribution (Figure 7.1b), which can be “binned” into a discrete distribution or even two states (e.g., either more or less than a particular value). Figure 7.1. Probability distributions.

70 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Probabilities are defined on a range from 0.0 (impossible) to 1.0 (guaranteed), so that the sum of probabilities of a comprehensive and mutually exclusive set of values must equal 1.0. For continuous distributions, the relative likelihood value is defined so that it integrates to 1.0. Uncertainties in combinations of factors are generally described by the probability distribution of each factor, in combination with a correlation coefficient, or by “con- ditional” distributions (Figure 7.1c). Performance Measures As discussed in Chapter 3, several key performance measures are of interest for rapid renewal projects: schedule, cost, and disruption through construction, longevity after construction, and combined performance. • Schedule. Key milestone dates (e.g., start of operations) or durations (e.g., time to replacement) typically are of interest. The entire schedule can be modeled via critical path analysis, in which (1) a complete and nonoverlapping set of project activities is identified; (2) their sequence (in terms of precedence requirements) is identified (e.g., visually in a flowchart); (3) activity durations, lags, and/or external milestone dates are assessed; and (4) early start and end dates are determined for each activity, which defines the critical path (and float for non–critical path) activi- ties and critical milestones and durations of interest. • Cost. Inflated costs through specific milestones (e.g., through construction) typi- cally are of interest. Costs can be modeled as follows: (1) a complete and non- overlapping set of project cost items is identified; (2) quantities and uninflated unit costs (including appropriate markups) are assessed for each item, consistent with the schedule (e.g., for overheads); (3) uninflated costs are determined for each item by multiplying the quantities and uninflated unit costs; and (4) inflated costs are then determined depending on when the various cost items occur (schedule of project activities and their relationship to the cost items) and on relevant inflation rates. The various cost items can be allocated to the project activities (e.g., 60% to Activity x and 40% to Activity y) to generate a cost-loaded schedule, and variable inflation rates for specific activities can be used. • Disruption. This is defined in terms of equivalent lost user person-hours, which includes traffic delays and detours, as well as business and other socioeconomic impacts. Disruption is assumed to be approximately additive, and thus can be modeled as follows (as discussed in Chapter 4): (1) a complete and nonoverlapping set of disruptive activities is identified; (2) the average disruption rate and duration for each activity are assessed, where the disruption rate might be determined on the basis of assessments of the average delay per person and average number of people affected per day; (3) the disruption is then determined for each activity by multiplying the average delay per person for that activity, the average number of people affected per day during that activity, and the duration of that activity; and

71 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS (4) the schedule of disruption can then be determined (if desired) by identifying when the disruptive activities will occur (e.g., according to the schedule activities). • Longevity. This is defined (see also Chapter 4) as the net present value (NPV) of costs and disruption (translated to equivalent cost) for operations and maintenance (O&M) and replacement, considering schedule (time to replacement) and using an appropriate net discount rate (which is a DOT policy issue rather than a technical one). The objective is to minimize this NPV. In this way, difficult (expensive or disruptive) O&M or replacement, or a short time to replacement, will be appro- priately “penalized.” Hence, longevity can be modeled as follows (as previously discussed in Chapter 4): (1) the average uninflated cost and disruption associated with O&M (e.g., on an annual basis) and with replacement, and the duration of O&M to replacement, are assessed; (2) the net discount rate and trade-off value (cost equivalence) of disruption are established; and (3) the NPV of cost and dis- ruption is determined by translating annual O&M and replacement disruption into equivalent cost terms and then adding them to annual O&M and replacement cost, respectively, and then discounting annual and replacement equivalent cost to NPV and adding them together. • Combined. Severity is defined as a change in the combination of the above per- formance measures, considering trade-offs among them. Severity can be modeled as follows: (1) the change in each of the performance measures is determined, as discussed above; (2) the trade-off value of schedule (advancing the operations date), of disruption (decreasing lost person-hours), and of longevity (decreasing the NPV of O&M and replacement cost and disruption) is established; and (3) the change in equivalent cost is determined by summing (a) change in inflated cost, (b) product of change in operations date and value of schedule change, (c) product of change in disruption and value of disruption change, and (d) product of change in longevity and value of longevity change. Deterministic Versus Probabilistic Analysis Deterministic (or traditional) analysis calculates one set of outcome values for one set of input values. It typically ignores the uncertainty in those inputs and the result- ing uncertainty in the outcomes. Probabilistic analysis, on the other hand, calculates probability distributions for the outputs as a function of correlated probability dis- tributions for the inputs (see Figure 7.1). This is done in one of two ways: analytical solutions or Monte Carlo simulation. • Analytical solutions can be done in several ways: — A discrete probability distribution can be determined, as shown in this simple example for cost, for a set of representative scenarios. However, in most cases, there are too many scenarios to tractably represent all possible combinations in such a combinatorial “tree.”

72 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS — The mean value and standard deviation of the outputs can be determined ap- proximately as a function of the mean and standard deviation of each input (in conjunction with the correlation coefficients between each pair of inputs), for example, via first-order second-moment and related point-estimate meth- ods. Although such approximate solutions are relatively simple for the mean (i.e., the mean value of an output is simply the deterministic function of the mean values of the inputs), it becomes more difficult and even impractical for the standard deviation (especially when inputs are correlated and for nonlinear models, such as for schedule). Also, except for some special cases in which the form of the probability distribution can be assumed (e.g., the sum of a large number of independent variables is a Gaussian distribution, and similarly the product of a large number of independent variables is a lognormal distribu- tion), the entire probability distribution is not developed. Only its mean and standard deviation are developed, so specific percentiles cannot be determined without further assuming a distribution form for the output. • Monte Carlo simulation can approximate the entire probability distribution of each performance measure, as well as the sensitivity of each performance measure to the various inputs, as follows: 1. A large number of possible sets of inputs (each set with a known probability of occurring) are developed by sampling (either randomly or more focused) the various input probability distributions (appropriately considering their correlations). Example of Analytical Solution for Discrete Distribution Base and two risks (R1 and R2), each with cost impact 9 2014 01 13 R09 08 Guide Chapter 7_final for composition.docx2013.12.09 R09 08 Guide Chapter 7_final for composition.docx • Analytical solutions can be done in several ways: o A discrete probability distribution can be determined, as shown in this simple example for cost, for a set of representative scenarios. However, in most cases, there are too many scenarios to tractably represent all possible combinations in such a combinatorial “tree,” [Insert Box 7.2] o The mean value and standard deviation of the outputs can be determined approximately as a function of the mean and standard deviation of each input (in conjunction with the correlation coefficients between each pair of inputs), for example, via first-order second-moment and related point-estimate methods. Although such approximate solutions are relatively simple for the mean (i.e., the Example of Analytical Solution for Discrete Distribution: Base and two risks (R1 and R2), each with cost impact Base scenario (B) → $B Risk 1 occurs (R1) → $R1 Risk 1 does not occur (R1’) Risk 2 occurs (R2) → $R2 Risk 2 does not occur (R2’) Risk 2 occurs (R2) → $R2 Risk 2 does not occur (R2’) B∩R1∩R2 → $B+$R1+$R2P = P[R1] x P[R2|R1] B∩R1∩R2‘ → $B+$R1P = P[R1] x P[R2’|R1] B∩R1’∩R2 → $B+$R2P = P[R1’] x P[R2|R1’] B∩R1’∩R2‘ → $BP = P[R1’] x P[R2’|R1’]Where:P[x] is probability of x occurring P[x’] is probability of x not occurring = 1 – P[x] P[x|y] is probability of x occurring if y occurs $x is cost impact if x occurs

73 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS 2. A set of outputs is developed for each set of inputs, using the deterministic model. Each set of outputs has the same probability of occurring as its set of inputs. 3. The large number of possible outcomes for each performance measure, where each outcome has a known probability, is statistically analyzed to determine the probability distribution of that performance measure. This sampled population of outcomes is inferred to adequately represent the actual population of pos- sible outcomes. 4. Correlations among the performance measures, as well as between each perfor- mance measure and each input, can also be determined statistically. Non–Risk-Based Versus Risk-Based Analysis Risk analysis can be conducted with or without identifying and quantifying individual risks, which might or might not occur. • In a non–risk-based approach, project uncertainties are “lumped” or “rolled up” into allowances (or contingencies) that are applied at high levels within the analysis: — For deterministic analysis, these allowances are intended to reasonably cover the various uncertainties. For example, a contingency of 20% of the base con- struction cost might be considered appropriate (based on published guidance) at a particular point in project development. — For probabilistic analysis, uncertainties in specific items are assessed, implicitly combining base uncertainties and risks. For example, a factor can be applied to a base cost item to express the range of that item, from the base cost item at the 10th percentile to the factor times the base cost item at the 90th percentile. Such a factor can be assessed on the basis of judgment (which is very difficult to do accurately and defensibly at such a lumped level) or, if enough data are available for that base cost item (which is very unlikely), based on statistics, essentially averaging all of the projects included in the database (“one size fits all”). • On the other hand, risk-based approaches explicitly address individual risks that can affect particular project elements. Risk-based approaches allow for more de- tailed uncertainty analysis, considering the uniqueness of each project, and facili- tate formal risk management planning, and are the focus of this guide. Time-Variable Versus Time-Independent Analysis For processes that vary significantly with time, the element of time should be consid- ered in the risk analysis. For many applications, a “pseudo–time-based” modeling approach (e.g., through use of a project cost-loaded schedule model) can adequately capture the key time-dependent features of projects without explicitly modeling the passage of time. For example, seasonal delays, inflation, and extended overheads can all be adequately incorporated in the model, and cash flow (or, in reverse, contingency drawdown) can be calculated.

74 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Subjective Versus Objective Assessment of Input Information As discussed in Chapter 6, when an adequate database of information related to a par- ticular variable is available, an objective, or statistical, approach can be used to develop inputs to the risk analysis. However, when statistical information is not available, the opinion of experts can be elicited, de-biased (as discussed in Chapter 6), and quanti- fied in the form of subjective assessments. Because most transportation projects—and particularly rapid renewal projects—are relatively unique, adequate statistical infor- mation is generally not available, and properly obtained subjective assess ments are required to conduct risk analysis. Facilitated consensus among a broad group of ex- perts helps to enhance accuracy and defensibility of such assessments. Decoupled Versus Integrated Analysis It is possible to conduct risk analysis on various project performance measures (e.g., cost, schedule) separately from one another. However, typically such decoupled analy- ses either ignore important relationships between these measures or treat relationships in an ad hoc manner. Integrated analyses explicitly identify, quantify, and model re- lationships (correlations and dependencies) between input variables and output per- formance measures. For example, an integrated cost and schedule analysis explicitly models the various relationships between inflated project cost and schedule. Initial Versus Updated Analysis Risks as well as the base generally evolve over time as the project develops and status, conditions, and plans change and new information becomes available. Once signifi- cant changes have occurred, the previous analysis (and its results) becomes outdated and should be updated to stay relevant. For example, a risk analysis (“diagnosis”) is typically performed before risk management planning (“treatment”) to identify targets for risk management. Plans will then change, based on risk management planning, and the risk analysis should be updated to consider those new plans. Levels of Detail, Accuracy, Defensibility, and Effort The level of detail can vary from simple algorithms with few but independent inputs to complicated algorithms with many correlated inputs. Although too little detail gener- ally involves too much approximation, too much detail can introduce errors, as well as unnecessary effort. The level of accuracy is a function of the method of analysis and level of detail chosen, as well as the accuracy of the inputs. The level of defensibility is a function of (a) the level of consensus achieved on inputs and the credibility of those involved; (b) the method of analysis chosen, espe- cially its logic and transparency; and (c) documentation of how the assessments were elicited or derived and how the analysis was conducted. The level of effort is a function primarily of the method of analysis and level of detail chosen, and of the accuracy, documentation, and level of consensus achieved and experts involved. Hence, the requirements for the levels of accuracy and defensi- bility must be balanced with the level of effort required to achieve those requirements.

75 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS PROCESS OF RISK ANALYSIS The risk analysis process is relatively straightforward, consisting of the following eight steps, which are subsequently described in more detail: Step 1. Identify the desired outputs or types of results from the risk analysis. Step 2. Select an appropriate method or approach for conducting the risk analysis. Step 3. Define a model of the system (i.e., project development), which also defines the inputs and relates the inputs to the outputs. Step 4. Define a project base (exclusive of risks). Step 5. Identify risks and opportunities relative to that base. Step 6. Quantify the risk analysis inputs (both base and risk factors), including their uncertainties and correlations. Step 7. Implement the model with uncertain (and correlated) inputs to determine un- certainty in the desired outputs and the sensitivity of the outputs to the inputs. Step 8. Document, check, and update (as needed). These eight steps are discussed in more detail. Step 1. Identify the desired outputs or types of results from the risk analysis. It is important to identify and adequately but efficiently answer the right questions. A risk analysis that does not address the DOT’s key questions is of limited use. As previ- ously discussed generally in Chapter 2, the desired outputs typically involve specific aspects of the project’s performance measures, including • The project’s total inflated cost, key schedule milestones, and cash flow through construction, and especially for rapid renewal projects, disruption through con- struction and longevity. Specific aspects of these broad performance measures might also be of interest, for example, construction contract cost and duration. This might include uncertainty in those performance measures, to help determine appropriate budgets, milestones, and contingencies. • Sensitivity of specific performance measures (e.g., a combined performance mea- sure) to each of the inputs, especially risks, to help develop risk management plans and proper allocation of the risks in the contract. These desired outputs should not be constrained by “canned” software outputs, because methods are available that can produce virtually any type of output. The accu- racy and defensibility requirements for the results should be established, appropriately considering the level of effort required to achieve them. The following guidance regarding the project scope and the evaluation strategy applies: • Evaluate the entire project. Consider all project phases and elements, including maintenance and operation where applicable, as described in previous chapters.

76 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Be careful not to focus project risk assessments too narrowly on construction. This is a mistake because many of a project’s largest risks and other uncertainties can occur early in a project’s development. • Evaluate all relevant performance objectives. For rapid renewal projects, con- sider disruption during construction and longevity (i.e., postconstruction cost and disruptions as well as postconstruction schedule) along with cost and schedule through construction. • Identify all possibilities, but stay focused on the key issues. Make sure to con- sider all possible outcomes, but do not get bogged down on insignificant items. Do not artificially exclude any significant uncertainties (including risks and op- portunities) from the analysis, because ignoring or otherwise excluding signifi- cant uncertainties, risks, and correlations will yield results that underestimate the true uncertainties, and provide misleading or even incorrect results that will not stand the test of time. However, if the DOT wants conditional analysis of various scenarios to help them evaluate internal decisions (e.g., regarding pro- curement method), then the results should be clearly “qualified.” Step 2. Select an appropriate method or approach for conducting the risk analysis. An appropriate method must be selected to provide the desired types of results, as identified in Step 1. Also as discussed generally in Chapter 2, the appropriate method depends on the desired outputs. For example, for DOTs who want to establish project budgets and schedules with a quantified confidence level (e.g., 80% probability of success), as well as conduct risk management, a viable approach is probabilistic, risk- based, integrated cost and schedule modeling. However, if the DOT is only interested in quantifying project cost in current (uninflated) dollars, then there is no need to model project schedule (although there might be extended overheads). Similarly, if the DOT is only interested in project schedule, there is no need to model project cost. Typically, however, DOTs are interested in predicting both cost and schedule. Because inflated cost and schedule are functionally linked, DOTs should in this case conduct integrated (or joint) cost and schedule modeling. Furthermore, DOTs are often inter- ested in evaluating the likelihood that their existing budgets will be met or establishing a budget (or contingency) with a reasonable likelihood for success. When this is the case, probabilistic modeling (i.e., appropriately considering uncertainties, correlations, and probabilities) is appropriate. Moreover, if contingency drawdown is desired, then an integrated cost and schedule model (which models cash flow) is needed. As will subsequently be discussed, risk-based models are needed for determining sensitivity of performance measures to risk and opportunities. However, if the DOT wants to determine the sensitivity of the target percentile of a performance measure (e.g., escalated cost) to the various risks and opportunities and other uncertainties, then special analyses are required, although still based on the results of a probabilistic, risk-based, integrated cost and schedule model (Figure 7.2).

77 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Assuming a qualified modeler, DOTs can choose from several commercially avail- able software packages to perform probabilistic, risk-based, integrated cost and sched- ule modeling. A few canned packages also conduct risk-based analysis. Otherwise, a Microsoft Excel workbook, with a commercially available add-in to do Monte Carlo simulation, can be used. Step 3. Define a model of the system (i.e., project development), which also defines the inputs and relates the inputs to the outputs. For project risk analysis, a numerical model of the project’s cost and/or schedule must typically be developed to adequately but efficiently combine and transform specific inputs into the desired outputs, consistent with Steps 1 and 2. For example, cost- loadable scheduling software or a suitably structured spreadsheet is typically used as a model to calculate the project’s ultimate inflated cost and schedule. Such a spreadsheet can be expanded to include other performance measures (disruption and longevity), whereas scheduling software is generally not as flexible. Above all else, however, the numerical model must adequately represent the system (i.e., project development in this case) being modeled to avoid introducing significant model error that could pro- duce misleading results. For rapid renewal, the model should generally consist of the following linked ele- ments (as previously described): • Schedule. Calculate (via critical path analysis) early start and end dates, as well as float, of each flowchart activity based on either its precedence logic (including lags) Figure 7.2. Probabilistic risk-based integrated cost and schedule model. Figure 7.2 R09 Guide

78 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS and duration or, if no precedence requirement, its milestone date. Durations can be base-only (which might be uncertain) or base plus realized risks, which in turn as- sume partial overlap of delays if multiple risks are realized in a particular activity. • Cost. Calculate total unescalated cost by simply summing the costs of a compre- hensive and nonoverlapping set of cost items (i.e., the cost estimate). Calculate to- tal escalated cost by allocating the cost items to the various flowchart activities (via a matrix), creating a simple cost-loaded schedule, and then escalating the cost of each activity based on its midpoint (from the calculated schedule) and its assessed escalation rate, which might vary among activities and from year to year. Typi- cally, calculate total cost only through construction with post construction cost considered under longevity. Each cost item can be a base cost (which in turn can be calculated from an average unit cost and a quantity, either or both of which might be uncertain) or a realized risk cost (some of which might be triggered by a schedule delay). Both kinds of costs are assumed to be additive in a particular activity. Escalation rates might also be uncertain. • Disruption. Calculate total disruption by summing the disruption associated with each flowchart activity. Typically, calculate total disruption only through construc- tion with postconstruction disruption considered under longevity. The disruption associated with each flowchart activity can be a base value (which in turn can be calculated as the product of the duration of disruption, the average number of people affected by disruption per day, and the average delay per affected person, any of which might be uncertain) or base plus a realized risk value (which might be triggered by a schedule delay). Both kinds of value are assumed to be additive in a particular activity. • Longevity. Calculate the net present value (NPV) of postconstruction cost and dis- ruption, based on the unescalated cost and disruption associated with O&M and replacement, the calculated schedule of O&M and replacement, and the estab- lished net discount rate and value of disruption (see Chapter 4). The unescalated costs and disruption for each activity can be base-only (which might be uncertain) or base plus realized risks (as discussed above). • Combined. Calculate the total equivalent escalated cost of the project, by translat- ing (via trade-offs) disruption through construction, construction completion date, and longevity into equivalent escalated cost and summing with the total escalated cost through construction (see Chapter 4). These can be base-only (which might be uncertain) or base plus realized risks. Step 4. Define a project base (exclusive of risks). The project base must be defined consistent with Steps 1–3, as described generally in Chapter 4. As noted in Step 1, this might include alternative scenarios (e.g., represent- ing internal decisions) for which conditional analyses are conducted to help make those decisions.

79 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Step 5. Identify risks and opportunities relative to that base. A comprehensive and nonoverlapping set of project risks and opportunities must be identified consistent with Steps 1–4, as described generally in Chapter 5. These risks and opportunities are relative to the base (Step 4). Step 6. Quantify the risk analysis inputs (both base and risk factors), including their uncertainties and correlations. The various risk analysis inputs must be adequately but efficiently assessed consistent with Steps 1–5, and as described generally in Chapter 6. These risk analysis inputs include • Base factors, including the base uninflated direct cost of each activity (or more detailed factors such as quantities and unit costs of various items, and their allo- cation to activities), activity base duration, lags, milestone dates, activity base dis- ruption, and base escalation rates for each type of activity; • Each impact scenario (in terms of quantitative changes in uninflated direct cost, duration, and disruption by activity) and its probability of occurring; and • Other policy factors, including postconstruction discount rates, value of disrup- tion, value of schedule, and value of longevity. If quantification of uncertainty in performance measures is desired, then the uncer- tainties in (and correlations among) these risk analysis inputs must be assessed. Among all the steps in risk analysis, quantifying uncertain inputs is perhaps the most problematic, because unqualified personnel can easily miss or improperly assess uncertainties and correlations. Therefore, DOTs should ensure that only qualified staff (with formal probabilistic training and relevant experience) attempt to quantify proba- bilistic inputs. As stated previously, only limited guidance on how to conduct quantita- tive risk analysis is provided in this guide because the topic is so expansive and several good references are available for probability theory and probabilistic and uncertainty analysis (see References section). However, some key guidance for quantifying uncer- tainty, which typically is not highlighted in common references, is provided here: • Variable definition. The variable being assessed should be clearly defined, so that everyone has a common understanding. Errors in input assessments, or their sub- sequent misuse, and difficulties in achieving consensus on such input assessments often arise from such misunderstandings. For example, the uncertainty in a value on any particular day, where that value changes significantly from day to day (“variability”), is very different from the uncertainty in the average value over all the days of interest (“ignorance”), which might be the intent and how the value is actually used in the analysis. In other words, there is a significant difference be- tween variability and ignorance, which should be recognized: uncertainty due to ignorance can be reduced by additional information, whereas variability cannot. Hence, the model will define the variable, and whether variability or ignorance is the main source of uncertainty.

80 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS • Distribution. For significant factors (i.e., those that can greatly affect the outputs), the full range of possibilities and their relative likelihoods should be assessed: — When the range of possibilities is continuous (e.g., a cost change of anywhere from $1 million to $2 million), a continuous probability distribution (as illus- trated in Figure 7.1) should be used. To develop this distribution, reasonable lower and upper limits (bounds) should be identified first, and then intermedi- ate values and their relative likelihoods should be addressed. The most likely or mean values should not be focused on first, because this will tend to lead to underestimation of the actual bounds and, therefore, of uncertainty. If low values are preferable (e.g., costs), then the reasonable lower bound represents a very optimistic value and the reasonable upper bound represents a very pes- simistic value; conversely, if high values are preferable (e.g., benefits), then vice versa. The level of conservatism associated with these bounds should be clearly established beforehand; for example, it is typically specified (based on research) that the reasonable lower bound corresponds to the 10th percentile (for which there is a 10% chance that the actual value will be less than that and a 90% chance that the actual value will be greater than that) and the reasonable upper bound corresponds to the 90th percentile (for which there is a 90% chance that the actual value will be less than that and a 10% chance that the actual value will be greater than that), so that there is an 80% (4:5) chance of being within this range. Some training of the assessors might be required to ensure that they understand what 10% chance means (e.g., by identifying common events that have a 10% chance of occurrence for comparison). A common probability dis- tribution form (e.g., a normal or Gaussian distribution) is then fitted to the range and other percentiles, based on judgment regarding the shape of the dis- tribution (e.g., symmetry, tails). However, there should not be a constraint of using only particular probability distributions (e.g., because they are conve- nient). Uncertain inputs should be quantified with reasonable representations of the relative likelihood for the various outcomes, and in particular should reflect the uncertainty as envisioned by the experts making the assessments. — When the range of possibilities is discrete (e.g., the risk either occurs or does not) or based on outcomes from potential scenarios (e.g., the DOT builds either a bridge crossing, or a tunnel crossing, or an at-grade crossing), consider using a discrete probability distribution (as illustrated in Figure 7.1) or an event tree (as illustrated in a previous example) to appropriately structure and quantify the risk. In the case of a comprehensive and mutually exclusive discrete set of possibilities, it is useful to first rank the possibilities (x is more likely than y) and then assess their relative differences (x is twice as likely as y) to determine their probabilities (recognizing that the probabilities must sum to 1.0).

81 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Conversely, for relatively insignificant factors, only the mean value (instead of the full range of possibilities) is generally needed. Assessing their full range of possible values would not significantly affect the results, but would take significant effort, and would thus not be cost-effective. • Correlations and dependencies. As previously discussed, a probability distribu- tion expresses the uncertainty in the value of a particular factor (either input or output). However, the uncertainty in the complete set of factors (especially input factors) is generally needed. Some factors might be related (e.g., because of a com- mon underlying factor), such that if one factor x is on the high end of its range, the related factor y would also tend to be on the high end of its range (positive correlation) or on the low end of its range (negative correlation). Some factors might be a function of (“conditional on”) other factors (e.g., the probability of Event B occurring might be different if Event A happens or not). Such relationships can be expressed in terms of a correlation coefficient for continuous or discrete distributions, or in terms of independent and dependent variables, in which the dependent variable has a conditional probability distribution that is a function of the value of the independent variable. These relationships among uncertain input factors should be adequately assessed and subsequently incorporated in the analysis. Otherwise, the uncertainties in the outputs will not be correctly deter- mined, typically being underestimated if such relationships are ignored (as subse- quently discussed). However, correlations among factors that are described only by their mean value (as opposed to a distribution) do not need to be assessed. Also, dependencies among events (as described by conditional probabilities) can often be avoided by combining these related events into a set of scenarios, each of which has a probability of occurrence (e.g., probability of Event A and Event B occur- ring). Note that probability distributions for outputs are conditional on the prob- ability distributions used for the inputs, which in turn are conditional on various assumptions (including exclusions). If these assumptions turn out to be invalid, then the probability distributions for the inputs and thus the outputs might not be correct and could be misleading. • Subjective assessment. For factors that must be subjectively assessed (because a statistically valid data set is not available), judgment biases (both management and cognitive, as discussed in Chapter 6) on the part of the assessors can result in errors. However, such biases can and should be countered to the extent possible by qualified facilitators and by achieving consensus among a broad group of ex- perts, including those that are independent of the project. The assessments should be consistent with all available information, which will generally support some values as being more likely than others, and might even preclude some values. As discussed in Chapter 8, some key input uncertainties can generally be reduced by obtaining specific new information that reduces the degree of ignorance.

82 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Step 7a. Implement the model with uncertain or correlated inputs to determine uncertainty in the desired outputs. The model must be adequately but efficiently implemented consistent with Steps 1–6. For project risk analysis, this involves translating the various inputs (base factors and risk factors) into all the outputs of interest (project performance measures, such as cost, schedule, disruption and longevity), as previously discussed, but also includes For example: To determine the total unescalated project cost ($ ) from the unescalated costs ( ) of a comprehensive and nonoverlapping set of cost items (i): The mean of : m m The variance of : (normal bell-shaped curve) with: v v m v • iff are all perfectly positively correlated : • otherwise v v where p x is probability distribution of x m x is mean value of x v x is variance of x x is specific percentile value of x is standard normal probability function for specific percentile (%), where, for example, = 0.842 is covariance between and v v is a correlation coefficient between and $ $ $ $ $ $ % $ $ * $ % $ % $ $ $ 2 cov $ ,$ % cov $ ,$ $ $ ,$ $ $ $ ,$ T T T i T T i T T i T T T T T i T T i T T T T T T T T T T T all all all % all all % $Ti i i i i i i j i j j i j i j i j ∑ ∑ ∑ ∑ ∑ ∑∑ φ φ 80%φ ρ [ ] [ ] [ ] [ ] =   ≈     ≈     ≈  +     ≈     ≈   +     =         $Ti $T $T • iff are all independent, p[ ] is approximately “Gaussian” $Ti $T $Ti $Ti $Tj i = 1 to n j = i+1 to n

83 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS translating uncertainties in the inputs into uncertainties in the outputs. Several good, although technical, references are available on propagating uncertainty (see References section). However, as previously discussed, for project risk analysis, there are essen- tially two general ways to propagate input uncertainties through a model: analytical approaches and numerical approaches (such as Monte Carlo simulation). A simple example of an analytical solution is shown for unescalated cost, which is a simple linear model. Although such analytical solutions are often not tractable for other performance measures, especially those that are more complex and nonlinear, they do provide some insight. The results in the example at the end of this chapter, on the other hand, are based on Monte Carlo simulation. If performed properly, simulation is a convenient and appropriate way to propagate uncertainty (even for nonlinear models) and to con- duct project risk analysis. Simulation capability is available for most popular project cost and scheduling software packages, as well as for many modeling platforms (e.g., Microsoft Excel). Regardless of the modeling method used, it is important to adequately incorporate the correlations in inputs. As shown in the simple example, there are typically two extreme (bounding) cases for correlations: total independence and “perfect” positive correlation. The results, especially in the tails of the distribution, can be very dif- ferent for these two extreme cases, with the variance and higher percentiles much greater for perfect positive correlation. Generally, for appropriate correlations, the distribution will be between these two extreme cases, with the total independent case under estimating (sometimes significantly) the uncertainty and the perfect positive correlation case overestimating (sometimes significantly) the uncertainty. Analytical approaches can incorporate correlations among the input factors through more com- plicated equations. Monte Carlo simulation can appropriately incorporate correla- tions among the input factors during the process of sampling those input factors, so that appropriate combinations of input factors are generated and used to determine the output populations. Because model inputs can be correlated and because model outputs can be func- tionally related in the model (e.g., because of common inputs), the various outputs might be correlated. For example, a risk that has a cost and a schedule impact will affect both cost and schedule, so that these two outputs would be correlated because of this common factor. On a bigger scale, escalated cost is affected by schedule (i.e., the escalated cost increases with schedule increase), so that these two outputs will obviously be correlated. These correlations in outputs are important if the outputs will be combined (e.g., into an overall measure of performance), as has been suggested here, for the same reasons as discussed above (i.e., the uncertainty in that combined measure would be underestimated if such correlations are ignored). There are two primary ways to deal with this correlation: (a) determine the outputs separately, assess (e.g., subjectively) the correlation among those outputs, and incorporate those correla- tions in any analysis in which those outputs are combined; or (b) determine all out- puts jointly and combine them appropriately using an integrated model during Monte Carlo simulation (see Figure 7.2). Approach (b) is recommended.

84 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Step 7b. Determine the sensitivity of the outputs to the inputs. The results must be adequately but efficiently analyzed to determine the sensitivity of those results to the various input factors (e.g., to subsequently guide risk management, as discussed in Chapter 8). The traditional way of determining sensitivity is to change each input by a specific amount (e.g., zero out a risk) and to then recalculate the out- puts and measure their change (e.g., in the target percentile). However, this becomes quickly unmanageable, especially if the model involves Monte Carlo simulation. For- tunately, other approximate methods are available to do this more efficiently. For the previous example shown here, the sensitivity of various aspects (e.g., mean, variance, specific percentile) of an output (e.g., total unescalated project cost) to the various inputs (e.g., unescalated cost of each item) can be determined analytically for simple linear models, especially with independent inputs. For base factors, the contribution of their uncertainty to specific (“target”) percentile values can be determined by assuming that their variance goes to zero (i.e., Δv[$Ti] = −v[$Ti] in the simple example), with no change in the mean value. For risks, their contribution can be determined by assuming that both their mean value and their variance go to zero (i.e., Δm[$Ti] = −m[$Ti] and Δv[$Ti] = −v[$Ti] in the simple example), where • The mean value of a risk equals its probability of occurrence times its mean value if it occurs; and • The variance of a risk equals the sum of — Its probability of occurrence times the square of the difference between (a) its mean value if it occurs and (b) its mean value; and — One minus its probability of occurrence, times the square of its mean value. For example (see previous example): To determine the sensitivity of to each (one at a time) m m v v so that m v m v $ $ $ $ $ $ % $ $ * $ $ * $ % $ % $ T T T T T T T T T T T T T % % i i i i i i φ φ ∆ = ∆ ∆   = ∆   ∆   = ∆   ∆   ≈ ∆   + ∆   ≈ ∆   + ∆   ∆   ≈ ∆   T$ i$T • iff are all independent, p[ ] is approximately “Gaussian” with: $Ti $T • iff are all perfectly positively correlated :$Ti

85 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS For more complex nonlinear models, approximate linear models can be developed that use weights (actually first derivatives) for each input factor, where the weights are derived by regression analysis from the many results produced during Monte Carlo simulation. Then the sensitivity can be determined in the same way as described above. This is how the example at the end of this chapter was developed, in which the con- tribution of each of the many uncertain factors to the target percentile (80%) of total escalated cost was determined, with one particular risk identified as being most impor- tant on that basis. The sum of the changes in mean value associated with each risk will equal the change in the mean value associated with all the risks collectively, whereas the sum of the changes in a specific percentile (e.g., 80th) associated with each risk generally will not equal the change in that percentile value associated with all the risks collectively. Step 8. Document, check, and update (as needed). Each step in the above process should be adequately but efficiently documented, reviewed, and checked. In particular, another qualified person should review the model logic, inputs, and results to ensure that the results are accurate and appropriate. Sub- sequently, as inputs change, their assessments, and the analysis, should be updated. This process is often iterative, especially updating Steps 4–8 as a project evolves over time and the risks, as well as the base (especially uncertainty), change with chang- ing status, plans, conditions, and information. For example, after an initial analysis has been conducted to identify the key risks, risk management planning is conducted to proactively reduce those risks, albeit often at some cost (see Chapter 8). Hence, for a particular risk management plan, the risks as well as the base will have changed, so that the risk analysis should be updated, presumably (if the risk management plan is cost-effective) resulting in better predicted performance and lower contingency requirements. The forms (Figure 7.3) and Microsoft Excel workbook template (template and related training materials are available online at www.trb.org/Main/Blurbs.168369. aspx) that were previously referenced for structuring in Chapter 4 and for risk assess- ment in Chapter 6 have been developed to facilitate limited risk analysis for relatively simple projects (see Appendix C). The template incorporates appropriate models to automatically and adequately determine • The relevant mean base project performance measures as a function of specific mean base factors, as input on the project structure form; • The mean changes in project performance measures, and thereby change in the mean combined performance measure (severity) for each risk, as a function of specific mean risk factors, as input on the risk assessment form; and • The relevant mean base + risk project performance measures as a function of spe- cific mean base and risk factors, as input on the project structure and risk assess- ment forms, respectively.

86 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Figure 7.3. Forms and template (see Appendix C). Figure 7.3 for R09 Guide

87 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Although these models are deterministic, if mean values are used for inputs, then the models produce reasonable approximations of the mean values of the outputs. More sophisticated analyses, typically using Monte Carlo simulation in conjunction with these (or more complicated) deterministic models and uncertain model inputs, are required to determine full uncertainty in the performance measures. CONCLUSIONS ON RISK ANALYSIS Risk analysis is a valuable (but not absolutely necessary) element of the overall risk management process. The primary objective of risk analysis is to quantify a project’s performance measures, including its uncertainty. Quantifying a project’s performance measures enables project decision makers to make better decisions among project al- ternatives or for the selected alternative, to establish (or determine confidence in pre- established) budgets and milestones, as well as to quantitatively determine the severity of each risk with respect to that set of project performance objectives, which allows for better risk management planning. If the DOT plans to conduct risk analysis, which involves quantitatively assessing the inputs (and their uncertainties, including correlations) and developing a model to calculate the outputs (and their uncertainties, including correlations), it should select the best method for its particular application, and then be sure to have adequately trained personnel conduct the analysis to avoid common pitfalls. If conducted and interpreted properly, the results can provide the DOT with valuable insight into poten- tial future project performance. However, if not conducted or interpreted properly, the results can be misleading.

88 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS Example The hypothetical QDOT case study (see Appendix D), which is used to illustrate the various steps of the risk management process and includes a risk management plan (RMP; see Appendix E), involved using the principles and process outlined in this chapter, as documented in RMP Addendum X (Appendix E) and summarized below. QDOT used the mean base and unmitigated risk assessments to determine (using the Microsoft Excel workbook template) the approximate mean unmitigated project performance (i.e., schedule, uninflated and inflated cost, and disruption, both total for the project and by project activity) in the same way as for base project performance. Although these results were very approximate (because of simplifications in the analysis), it provided insight into the collective effect of the risks, before any additional mitigation. This information and these tools were also used to determine the mean severity of each risk, in terms of how much the combined performance measure is af- fected by that risk. Subsequently, a quantitative risk analysis was conducted (see Appendix E, RMP, Addendum X for inputs and results), for which: • A more detailed flowchart was developed (by consensus) by the facilitated group (see below). • Uncertainties in the unmitigated base cost estimate and schedule were assessed (by consensus) by the facili- tated group; for example, bridge structure cost ranges (10th to 90th percentile) from −20% to +20%, and is moderately correlated (coefficient of 0.75) with other construction cost items. • Unmitigated risk factor assessments were refined (by consensus) by the facilitated group (see below). • A more sophisticated probabilistic (via Monte Carlo simulation) integrated cost and schedule model was developed to represent the more detailed flowchart and implemented with the more refined unmitigated base and risk assessments. • Uncertainties in unmitigated project performance (i.e., project completion date and cost through construc- tion, both unescalated and escalated) were determined (see below). • Contributions of each risk and base uncertainty toward the target (80th percentile) escalated cost through construction and project completion date were determined (see below); for example, EP2 contributes $0.2 million to 80th percentile of escalated project cost, and ranks 13th. As will be discussed, the uncertainties in project performance can be used to determine appropriate budget, mile- stone, and contingency, and the sensitivity of the budget (not just the mean cost) to the various risks can be used to better guide risk management. (continued)

89 GUIDE FOR THE PROCESS OF MANAGING RISK ON RAPID RENEWAL PROJECTS 36 2014.01.13 R09 08 Guide Chapter 7_final for composition.docx QRA flowchart for QDOT US-555 and SH-111 Project Risk or Opportunity Probability of Occurrence (%) Cost Change if Occurs (2009 millions of $) Duration Change if Occurs (months) PD13 Change in environmental documentation Mutually exclusive scenarios: A. 50 (base) B. 40 C. 8 D. 2 A. 0 (base) B. +0.1 to Activity 2 C. +0.5 to Activity 2 D. +0.5 to Activity 2 and +1.0 to Activity 12 A. 0 (base) B. +1 to Activity 2 C. +6 to Activity 2 D. +6 to Activity 2 Quantitative Assessment for a Select Rapid Renewal Risk for QDOT US-555 and SR-111 Project Unmitigated project performance (cost) uncertainty and sensitivity of 80th percentile of escalated cost for QDOT US-555 and SH-111 Project ! Preliminary Design (to 30%) 1 Draf t Environmental Assessment (EA) 2 Prepare / Issue RFP 4 Remaining as of 12/1/2009 Finalize EA / Approval 3 DB Response / Review / Selection / Negotiate 8 Complete 136 months 6 months 6 months 2 months 6 months Notes: 1. Single Design/Build contract. 2. Advance Right-of-Way (ROW) Acquisition includes appraisals, offers, acquisition, relocation, and demolition for parcels that QDOT anticipates will be critical to early construction by the Design/Builder. 3. Advance Utility Relocations includes coordination, approvals, and relocations of utilities that QDOT anticipates will be crit ical to early construction by the Design/Builder. Additional relocations that might be required will be the responsibility of the Design/Builder during construction. Assumes minimal new ROW required for utility relocations. 4. QDOT will complete the Environmental Assessment (EA) and obtain all environmental permits before Notice to Proceed (NTP). 5. Construction duration includes typical winter shut-down period from November 15th through March 15th. 6. Construction includes construction permits, remaining utility relocations, and all construction -related effort. Remaining ROW acquisition by QDOT also occurs during this timeframe. QDOT’s US 555 / SH 111 Expansion Project Simplified Risk Assessment Flow Chart December 1, 2009 Rapid Renewal Delivery / Schedule Base Schedule (excluding risk): • Pre-Construction (up to NTP): 18 months • Construction (af ter NTP): 17 months • Total duration: 35 months Environmental Permitting 7 6 months Notice to Proceed 10 Funding 9 12 months Advance ROW Acquisition 6 9 months Advance Utility Relocations 5 Design/Builder Design 11 Design/Builder Construction 12 16 months 6 months S + 1 month 6 months remain 14 months remain Base date: 6/1/2011 Base date: 11/1/2012 VERSION 2: CONSERVATIVE PRE-CONSTRUCTION 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 10 15 20 25 30 35 Total Project Cost (millions) C um ul at iv e Pr ob ab ili ty (P er ce nt ile , C on fid en ce ) 2009 $ Year-of-Expenditure $! 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 CN12. Extended overheads as a function of project delays Compound Construction Inflation to Midpoint of Construction CP2. Uncertain D/B contracting market conditions at time of bid PD11. Cannot use City sewer system for project runoff (or City charges for use) RU3. Unwilling sellers SC6. Provide new lighting throughout project Identified Minor Risks (aggregate) Unidentified Risks (aggregate) RU8. QDOT helps City pay for water and sewer-line relocation RU2. Accelerating pace of development in interchange area Traffic Control (at 7% of subtotal A + Mob) CN2. Additional Maintenance of Traffic required EP2. Change in environmental documentation R2. Accelerating pace of development in interchange area EP1. Structures impacted by Main Street realignment are eligible for Historic Register CN3. Problems with planned accelerated bridge construction (ABC) technique PD5. Shoulders required on US 555 EP6. Additional wetland mitigation required for planned alignment PD3. Change configuration of SH 111 / US 555 interchange PD6. Shoulders required on SH 111 Contribution to the 80th Percentile Total Cost (YOE $M)

Next: 8 RISK MANAGEMENT PLANNING »
Guide for the Process of Managing Risk on Rapid Renewal Projects Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s second Strategic Highway Research Program (SHRP 2) S2-R09-RW-2: Guide for the Process of Managing Risk on Rapid Renewal Projects describes a formal and structured risk management approach specifically for rapid renewal design and construction projects that is designed to help adequately and efficiently anticipate, evaluate, and address unexpected problems or “risks” before they occur.

In addition to the report, the project developed three electronic tools to assist with successfully implementing the guide:

• The rapid renewal risk management planning template will assist users with working through the overall risk management process.

• The hypothetical project using risk management planning template employs sample data to help provide an example to users about how to use the rapid renewal risk management template

• The user’s guide for risk management planning template will provide further instructions to users who use the rapid renewal risk management template

Renewal Project R09 also produced a PowerPoint presentation on risk management planning.

Disclaimer: This software is offered as is, without warranty or promise of support of any kind either expressed or implied. Under no circumstance will the National Academy of Sciences or the Transportation Research Board (collectively "TRB") be liable for any loss or damage caused by the installation or operation of this product. TRB makes no representation or warranty of any kind, expressed or implied, in fact or in law, including without limitation, the warranty of merchantability or the warranty of fitness for a particular purpose, and shall not in any case be liable for any consequential or special damages.

Errata: When this prepublication was released on February 14, 2013, the PDF did not include the appendices to the report. As of February 27, 2013, that error has been corrected.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!