Guidelines for Managing Geotechnical Risks in Design–Build Projects (2018)

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.

28 This chapter focuses on the quantitative analysis of geotechnical risk for the typical design– build project. It provides methodologies for computing the magnitude of the risks identified using the tools found in Chapters 2 and 3. Risk quantification or measurement was shown in Stage 2 of the risk assessment process discussed in Chapter 1 (see Figure 1.1). 4.1 Risk Quantification The quantitative approach is a more involved process compared with the qualitative approach. This approach is usually reserved for larger, more complex projects. As an example, Washington State DOT recommends a quantitative cost risk assessment to be conducted for projects larger than $10 million and the use of a formal analysis approach for projects costing more than $25 million. There are several mathematical approaches for calculation of total risk impact, including analytical methods and Monte Carlo simulation techniques. Detailed descriptions of these methods are in DOT manuals and published books and reports. These recommendations are for the full risk assessment (not only for geotechnical risks); the NCHRP survey and inter- views showed that no DOT does exclusive geotechnical risk assessment. All the geotechnical risk assessment activity is conducted as part of the overall project risks. The purpose of risk assessment described in this report is to provide guidelines for quantifying the risk of geotechnical factors on a project’s cost and schedule. As such, the same general methodology for whole-project risk assessment can be used, but with emphasis on geotechnical risks. The following sections describe the quantification process and cover two alternative methods: (1) a non-simulation approach and (2) a Monte Carlo simulation-based approach. 4.2 Risk Register The basic tools are still the same. As was the case with the qualitative approach, the risk assess- ment team should consist of persons involved with the geotechnical and geoenvironmental design of the project. The team starts with the list of risks that have been identified (see Chapter 2). The most significant risks are entered into a risk register. The risk register is simply a table for docu- menting project risks and their characteristics. Several examples of risk registers are found on DOT websites. For each risk factor in the risk register, the team has to estimate the probability of occurrence (based on subjective knowledge of the geotechnical experts and possibly some statistics from similar past projects) and the impact of risk on project cost and schedule. The project cost and schedule impact may be estimated by a single value (deterministic) or with a range (probabilistic). The combined impact of all these risks is the total risk impact on the project. The process of information elicitation from the SMEs has been described in Chapter 3. C H A P T E R 4 Quantitative Geotechnical Risk Analysis

Quantitative Geotechnical Risk Analysis 29 4.3 Non-Simulation Approach In the non-simulation approach, the likelihood and the expected impact of the risk, if they occur, are estimated by the risk assessment team. The results are documented in the risk register. A sample risk register, taken from Virginia DOT (Form PM-103B) and shown in Figure 4.1 is suggested for this purpose. All the identified significant risks will be listed in this table. Each risk factor is entered in a new row. For each risk, the likelihood (probability of that risk event happening) should be estimated. The cost and schedule of the risk impact, in case it happens, need to be estimated and entered into the risk register. The same register can be used as a mitigation/ risk control tool. The mitigation strategy for each risk, the person responsible for mitigation, the cost of mitigation, and the likelihood of success should be entered into the risk register. RISK REGISTER Form PM-103 BRev. 01-27-2014 UPC # PROJECT LOCATION VDOT DISTRICT RISK ASSESSMENT RISK RESPONSE RESIDUAL RISK ASSESSMENT (AFTER APPLIED STRATEGY) Probability Impact Probability Impact Responsible Team Member Method Time/Budget Impact Probability Impact Probability Impact I. Roadway Design (Design Considerations; Design Exceptions, Status of Preliminary Plans, Typical Sections, Roadway Classification/Traffic Data) II. Bridge Design (Design Considerations; Staged Construction, Demolition of Existing Structure, Bridge Aesthetics) III. Right of Way (Total takes, Relocations, Railroad) IV. Environmental (Status of Environmental Document; Noise Study, Air Study, T&E, Water/E&S Permits, Hazmat, Stream Relocation) All Applicable Permits for E&S V. Utilities (Underground/Overhead Utilities; Power Lines, Gas Lines, Unknown (Governmental) Utilities) VI. Geotechnical (Preliminary Geotechnical Investigation Status, Boring Logs, Final Soils and Pavement Report, Minor Structure Foundation Design, etc.) VII. Drainage (Replacement/Flowable Fill of Culverts, Outfall, Large Pipe Installation) VIII. Construction (MOT; Time of Day Restrictions, Incentives/Disincentives, DBE, Staging Area; Clearing of Trees) IX. Public Involvement (Localities, Elected Officials, Citizens, Other Agencies) X. Approvals/Concurrence (VDOT/External Agencies) XI. Coordination with Other Ongoing Projects in the Corridor XII. Additional Issues (Third Party Requirements, Funding/Budget) PROJECT # Risks Comments/ Notes Mitigation Strategy Figure 4.1. Example risk register for quantitative risk assessment: DBE = disadvantaged business enterprise, E&S = erosion and sedimentation, MOT = maintenance of traffic, and T&E = threatened and endangered species [adapted from Form PM-103B, Virginia DOT (2015)].

30 Guidelines for Managing Geotechnical Risks in Design–Build Projects Such estimates are provided for all the risk factors. The SMEs who conduct the risk assessment in a workshop setting arrive at the estimated costs. Sometimes these values are completely subjective on the basis of an SME’s personal experience with previous projects and sometimes these costs are estimated on the basis of reasonable assumptions. Usually these assessments are done in a group setting, in which geotechnical experts review project documents to identify risks. One approach commonly used by the risk analysis team is to poll the experts and calculate the average value of the opinions. In general, the expected impact of risk can be estimated by multiplying the probability of occurrence by the risk impact (cost or delay) (Equation 4.1). This expected impact is in effect the average effect of the risk factor. Mathematically, these expected impacts (average impacts) can be summed to arrive at the total average cost and schedule impact. (4.1)EI P Ir r r= where EIr = expected impact of risk r, Pr = probability of risk r, and Ir = impact of risk r. An example of a hypothetical analysis is provided in Figure 4.2. Four geotechnical risk factors are listed and the probability of occurrence is estimated. For example, unidentified utilities are identi- fied as a risk factor and the likelihood encountering such risk is estimated as 50%. If such utilities are encountered, it is estimated they will add $250,000 to the project cost. The average impact of such a risk event is estimated as 50% × $250,000 = $125,000 as shown on the register. The total average effect of the four risk factors is estimated as $169,583. This approach will not establish the range of the potential cost increase as this figure only represents the average cost increase. The range of total cost or schedule impact can be calculated if information is available for the range of each risk factor. 4.4 Simulation Approach In the method described above, the risk impact was estimated with a single value. It is often difficult to estimate the impact accurately, given the conceptual nature of information at the time of geotechnical risk assessment. It is more reasonable that this impact can be estimated with a range. Figure 4.3 shows the process of risk quantification. Note that the cost and schedule impact are represented with a range rather than with a point estimate. A common approach is RISK REGISTER PROJECT # EXAMPLE Date No. Risk Factor RISK ASSESSMENT RISK RESPONSE Probability Impact Probability Impact Responsible Team Member Method Time/Budget Impact 1 Unidentified utilities causing delays during construction 50% $250,000 $125,000 2 Damage caused by differential settlements within bridge footprint after construction 20% $125,000 $25,000 3 Unknown groundwater flow direction 5% $58,333 $2,917 4 Damage to neighboring properties by heave due to excavation 10% $166,667 $16,667 TOTAL $169,583 Figure 4.2. Partially filled-out risk register with identified geotechnical risk factors.

Quantitative Geotechnical Risk Analysis 31 to try to bracket the cost or delay by providing a pessimistic and an optimistic value and then provide the most likely cost or delay. So each risk cost or delay is estimated with three estimates representing optimistic, most likely, and pessimistic scenarios. Ranges for risk cost or schedule are modeled as random variables using statistical distribu- tions. Various probability distributions have been used to estimate the cost and schedule impact of risks. The choice of the distribution depends on the agency’s preference and familiarity, pre- vious experience with similar projects, and mathematical convenience. Because the risks are quantified with ranges (and probability distributions) rather than with single estimates, estima- tion of their combined effect requires probabilistic methods. The combined effect of cost and schedule risks will be a statistical distribution (or range of outcomes). In its most simple form, the combined effect of costs is the sum of the costs of risk factors (Equation 4.2). . . . . . . . . . . . . . . . . . . . . . (4.2) 1 2 3 1 1 2 2 3 3 RC R R R p C p C p C = + + + = + + + In Equation 4.2, RC is the total risk cost, R1 , R2 , R3, and so on are the expected cost impact for the first, second, third, and so on risk factors, and p1 and C1 are likelihood and impact cost of the first risk factor, and so on. Each of these risk impacts is represented by a probability distribution (not necessarily the same type of distribution). It is often difficult to try to get to the distribution of RC in Equation 4.2 using a direct analytical approach. The mathematics of dealing with the products and sums of various distributions may become intractable. Because of this, a Monte Carlo simulation approach becomes a reasonable alternative, especially when correlations exist between various risk factors. A full discussion of statistical distributions is beyond the scope of this report but normal distribution, lognormal distribution, program and evaluation technique, and uniform or trian- gular distributions have been used for the purpose of risk assessment in transportation projects. 4.4.1 Monte Carlo Simulation Monte Carlo simulation is used to calculate the combined effect of geotechnical risk factors. In a Monte Carlo simulation, the modeler samples statistical distributions representing random variables (mainly cost or duration in risk assessment) using a computer. The simulation software generates random numbers for these costs (or durations) according to specified distributions and calculates the total cost (or duration). This process is repeated hundreds or thousands of times, each usually called an iteration or realization. The number of iterations depends on the confidence intervals desired for the results. It is understood that the larger the number of itera- tions, the more reliable the results. Each simulation iteration produces a single value for cost or delay caused by geotechnical risks. These values can then be organized into a histogram for total project cost (or duration). With this histogram, a probability density distribution and a cumula- tive distribution function for total risk cost (or duration) are compiled. These distributions can Identify all risk factors Estimate likelihood of risk event Calculate risk impact Provide a range for cost/schedule risk Figure 4.3. Risk quantification.

32 Guidelines for Managing Geotechnical Risks in Design–Build Projects then be used to decide on the next course of action and can also be used to determine reasonable amounts of contingency for the project’s budget or duration. 4.4.2 Example The hypothetical example of Figure 4.2 is used to illustrate a simulation-based risk assessment case. Four major geotechnical risk factors are identified and documented in the risk register. For the purpose of this example, the cost impact is modeled using triangular distributions with ranges elicited from SMEs (Figure 4.4). A triangular distribution is identified with three esti- mates for the minimum, most likely, and maximum values. As an example, the potential cost of the risk factor “Damage caused by differential settlements within bridge footprint after construc- tion” is modeled with a triangular distribution with the following values: • optimistic cost = $25,000 • most likely cost = $100,000 • pessimistic cost = $250,000 The likelihood is estimated as 20%, which is shown as a discrete distribution on the first row (probability of occurrence = 20% and probability of nonoccurrence = 80%). For each risk factor, the SMEs were asked to provide a range for cost and an estimate for the likelihood of the occurrence of the risk factor. For each data elicited from the SME, the facilitator should strive to get to a consensus regarding the estimate. In the absence of a consensus, an average of the expert opinions may be used as the input for the risk model. Commercial software is available (@RiskTM or @Crystal BallTM) that can perform Monte Carlo simulation. For this example, @Risk software was used to generate 10,000 iterations. For Figure 4.4. Input values for the risk factors.

Quantitative Geotechnical Risk Analysis 33 each iteration, the software sampled each of the input distributions, multiplied impact values by likelihoods, and summed up the outcomes to calculate the total risk. Figure 4.5 summarizes the outcome of these 10,000 simulation iterations. Figure 4.5 shows that the total geotechnical risks can have a cost impact ranging from $0.00 (in the case that none of the four risk factors materializes) to $740,134 with the most likely cost impact being $170,147, the expected cost if project risks are not mitigated. Figure 4.6 shows a cumulative distribution for the costs. If the agency is interested in setting up a contingency exclusively for geotechnical risks, then a $300,000 contingency will provide a confidence level of 80% against geotechnical risks. Alternatively, this analysis can be used to Most Likely Impact Va lu es x 1 05 Figure 4.5. Total cost of geotechnical risks. Figure 4.6. Cumulative distribution function for total cost of geotechnical risks.

34 Guidelines for Managing Geotechnical Risks in Design–Build Projects decide the level of premium that the owner may be willing to pay to transfer all DSC risks to the DB contractor. This analysis can also be used as a tool to assess the necessity of further geotechnical investiga- tions before going to bid. If the magnitude of risk is beyond the risk tolerance of the agency, then it makes sense to develop the geotechnical risk further and clarify the risk picture. The project is classified as high risk and if the letting of the DB project as a progressive DB is permitted by law, then it could provide a reasonable approach to deal with geotechnical risks. DB approach can be a suitable approach for the observational method of geotechnical design. The agency can benefit from the fact that designer and the contractor can work together to find effective solutions to geotechnical issues that are unearthed during the construction phase.