Byung-Cheol Kim

Probabilistic Forecasting of Project Duration Using Kalman Filter and the Earned Value Method

The earned value method (EVM) is recognized as a viable method for evaluating and forecasting project cost performance. However, its application to schedule performance forecasting has been limited due to poor accuracy in predicting project durations. Recently, several EVM-based schedule forecasting methods were introduced. However, these are still deterministic and have large prediction errors early in the project due to small sample size. In this paper, a new forecasting method is developed based on Kalman filter and the earned schedule method. The Kalman filter forecasting method (KFFM) provides probabilistic predictions of project duration at completion and can be used from the beginning of a project without significant loss of accuracy. KFFM has been programmed in an add-in for Microsoft Excel and it can be implemented on all kinds of projects monitored by EVM or any other S-curve approach. Applications on two real projects are presented here to demonstrate the advantages of KFFM in extracting additional information from data about the status, trend, and future project schedule performance and associated risks.

Combination of Project Cost Forecasts in Earned Value Management

Reliable cost estimates are essential for effective project control and the management of cash flows within the project and at the company level. Conventional approaches to project cost forecasting, which rely on detailed information developed for a specific project (the bottom-up estimate or inside view), often result in cost overruns. It is argued here that the inside-view project cost estimates should be adjusted by combining them with the outside (or top-down) view of the project, which is based on statistical models of historical project data. This paper presents a probabilistic cost forecasting method and a framework for an adaptive combination of the inside view and the outside view forecasts of project cost using Bayesian inference and the Bayesian model averaging technique. During the project execution phase, the Bayesian adaptive forecasting method incorporates into the predictions the actual performance data from earned value management and revises preproject cost estimates, making full use of ...

Sensitivity of Earned Value Schedule Forecasting to S-Curve Patterns

AbstractThis paper examines sensitivity of the performance of seven project duration forecasting methods in the earned value management (EVM) literature to characteristic patterns of planned value and earned value S-curves. Specifically, this paper aims at identifying relative robustness and early warning capacity of six deterministic methods and one probabilistic method with respect to the nonlinearity of progress curves and the schedule delay patterns. The sensitivity analysis in this paper shows that forecast accuracy and early warning credibility of deterministic methods are very sensitive to the S-curve patterns, especially early in a project. The results also indicate that the probabilistic method (the Kalman filter earned value method) is the only method among the seven alternatives that is robust with respect to the progress curve nonlinearity and the schedule delay patterns. Consequently, this paper would positively contribute to the practice of project schedule control by providing practical gui...

Integrating Risk Assessment and Actual Performance for Probabilistic Project Cost Forecasting: A Second Moment Bayesian Model

Forecasting the actual cost to complete a project is a critical challenge of project management, particularly for data-driven decision making in contingency control, cash flow analysis, and timely project financing. This paper presents a Bayesian project cost forecasting model that adaptively integrates preproject cost risk assessment and actual performance data into a range of possible project costs at a chosen confidence level. The second moment Bayesian (SMB) model brings more realism into project cost forecasting by explicitly accounting for inherent variability of cost performance, correlation between aggregated past and future performance, and the fraction of project completed at the time of forecasting. Functionally, the SMB model fully encompasses, as restrictive cases, two most commonly used index-based cost forecasting techniques in earned value management. The SMB model provides computationally efficient algebraic formulas to conduct robust probabilistic forecasting without additional burden of data collection or sophisticated statistical analysis. Numerical examples and simulation experiments are presented to demonstrate the predictive efficacy and practical applicability of the SMB in real project environments.

Dynamic Control Thresholds for Consistent Earned Value Analysis and Reliable Early Warning

AbstractEarned value management (EVM) facilitates practicing management-by-exception by triggering control actions only when they are needed. The efficiency and effectiveness of project control actions are influenced by the control thresholds used to identify meaningful performance deviations or trends from the plan. This paper presents a quantitative method for establishing dynamic control thresholds (DCTs) that change at different percent completion points according to the degree of achievability of the project objectives in time and money. The DCTs in this paper overcome two limitations of conventional fixed control thresholds. First, the DCTs provide consistent measures of risk and early warning signals for different EVM performance measurements and forecasts. Second, the DCTs provide additional information to detect false warnings from deterministic performance analysis and forecasting monitored by fixed control thresholds, especially early in a project. This paper mainly focuses on schedule performa...

Probabilistic Evaluation of Cost Performance Stability in Earned Value Management

AbstractCredibility of cost performance analyses and forecasts for effective control of construction projects depends on the stability of the cost performance index (CPI) as a leading indicator of future performance. This paper presents a simulation-based analytical model that evaluates the probability that the CPI of an individual project becomes stable at a particular time during the project’s duration. The CPI stability model is used to identify four cost stability driving factors, and their compounding effects are examined by a parametric study. A real-project application demonstrates that the CPI stability is an indicator of the overall project performance risk and thus, the specific conditions of each project must be properly accounted for in a CPI stability assessment. This paper contributes to enhancing the credibility of all CPI-based analyses in earned value management by providing insights into the underlying factors of cost performance stability and an analytical tool for assessing the CPI sta...

A Second Moment Approach to Probabilistic IRR Using Taylor Series

This article presents a practical approach for computing the internal rate of return (IRR) of stochastic cash flows. The mean and variance of the distribution of the IRR are derived by a second-moment approach using the means, variances, and correlations of the costs and returns in a cash layout. Programmed as an add-in function, the second-moment algorithm was applied to three examples in the literature and the accuracy of the second-moment method was compared with more sophisticated, computationally intensive methods. The comparison indicates that the second-moment approach can serve as a quick and viable tool for probabilistic IRR analysis.

Cost Performance as a Stochastic Process: EAC Projection by Markov Chain Simulation

AbstractEarned value analysis (EVA) has been widely used in the construction industry for cost prediction at completion. The EVA’s accuracy of early cost projections is low since the method assumes static cost performance during construction. A project’s cost performance is evidenced as a stochastic process. In an effort to improve the EVA’s accuracy of early cost predictions, this work reports a modified method of Markovian simulation cost projection (MSCP). Based on Markov chain simulation, MSCP simulates the probability distribution of the cost performance indicators for each period of a project, and predicts the final cost using the summation of each simulated period cost. The MSCP method is demonstrated and validated through a case study of a real-world power plant project. Data analysis indicates that MSCP improves the prediction accuracy four times higher than EVA. Findings also suggest that MSCP is able to capture erratic changes of cost performance throughout a project’s lifecycle and thus provid...

Probability Distribution of the Project Payback Period Using the Equivalent Cash Flow Decomposition

For stochastic cash flows, probabilistic approaches to determine a complete distribution of payback period are very limited. The payback analysis based on the net present value (NPV) has several advantages. For annual cash flows, however, the NPV-based method does not provide a complete payback distribution. This article proposes a new technique, the equivalent cash flow decomposition (ECFD), which converts an annual cash flow into an equivalent subannual cash flow at a desired level of precision. The ECFD technique can be used in conjunction with any probabilistic cash flow technique. This article demonstrates that the ECFD technique overcomes the discontinuity limitation of the conventional NPV-based payback period method and generates a complete distribution of the payback period of annual cash flows. Examples indicate that the proposed method is robust with the accuracy comparable to Monte Carlo simulation.

Probabilistic Performance Risk Evaluation of Infrastructure Projects

Forecasting is a critical function of project control and management. Reliable forecasting enables the project manager to make better informed decisions for timely control actions to prevent or mitigate adverse project outcomes, especially schedule delays and/or cost overruns. Recently, a new probabilistic method for project schedule forecasting was developed based on the Kalman filter method and the earned value method. In this paper, the Kalman filter forecasting method for schedule is extended to formulate a consistent and practical method for project schedule and cost performance forecasting. A numerical example is presented to demonstrate how the new method can be efficiently employed in real projects. Monte Carlo simulation is also conducted to evaluate the accuracy of the proposed method.