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.
CHAPTER 2: RESEARCH APPROACH The research approach can be broadly categorized into three phases: (1) Understanding the problem, (2) developing cost modeling approach, (3) data collection, (4) implementing the approach in a cost model, and (5) validating and refining the model. Several of the activities within these broad phases occurred in parallel as, understanding of the problem, the approach, and the framework of the cost model matured together. Understanding the problem As described above, there is currently a lack of standards in establishing cost estimate for the capital planning phase. The resulting cost estimates are therefore likely to result in large variations with actual costs. This can result in cancellations of projects or inefficient use of limited airport funding. The problem is particularly challenging since many small and regional airports do not have in-house staff dedicated to cost estimating. Other ancillary issues further complicate the problem. In particular, the use of contingency factors adds another layer of uncertainty to the estimate, especially since there are no established standards for their application. In order to better understand the scope and the nature of the research problem, the Research Team completed extensive background research. This effort included a literature review, identifying and documenting best practices for cost estimating in general and as applied to airport construction specifically. It also identified and reviewed existing cost estimating models. Another focus of the background research was stakeholder outreach. The Research Team conducted a stakeholder survey to identify existing practices and cost estimating needs. Developing the cost estimating approach The objective of this research project is to develop an interactive construction cost estimating model and associated database for airport capital projects, along with a guidebook documenting its use and best practices. The model should cover common airport construction projects, both in the horizontal and vertical domains. The model should make use of existing databases and take into account regional cost factors and inflation. The model should be flexible in its use, for example, by allowing for database updates and the ability to generate reports in Excel, PDF, and other formats. The proposed approach is to use a parametric cost estimating technique in which costs are correlated to observed (i.e., historical) data from actual construction projects. In this approach, multivariable regression analysis is used to model cost through mathematical functions known as Cost Estimating Relationships (CERs). The CERs model cost as a function of key cost drivers known as Candidate Independent Variables (CIVs). The variables are considered candidates because they are selected using subject matter expert input and are then tested for statistical validity and reasonableness. 3
The output of the model is a cost estimate for a single project or a portfolio of projects, with both a point estimate and a high/low range that represents risk. The costs are inflation adjusted and take into account regional variations. The inputs to the model that are necessary to derive these costs are the cost drivers represented by a set of CIVs that is unique to each type of project. The CIVs are the independent variables in the CERs, which represent the analytical component of the model. Data collection Two separate data collection templates were developed, one for horizontal and one for vertical construction projects. The templates matched the structure of the cost estimating database, by including a series of sub-templates, one for each project type. For each historical observation, fields for basic descriptive information were provided, such as a project description, location, and year of completion. Other data fields were used to store values for the dependent variable (i.e. construction costs) and all independent variables (i.e. CIVs) required by the proposed CER for the project type in question. Due to lack of usable data in electronic format, the data collection was conducted into two phases: an initial data collection effort, followed by targeted, supplemental effort. The results of the two data collection efforts are described in more detail below. Developing the cost model After reviewing best practices, the Research Team identified several guidelines for the model, including that it should: ⢠Be implemented in Microsoft Excel, or similar. The model should not require proprietary software and should be based on a platform that is ubiquitous and likely to be in use at most airports (e.g. Microsoft Office). ⢠Provide a guided approach in which the user is prompted to select pre-defined choices and/or enter required data along a logical path. ⢠Combine pre-defined data populated in the model by the Research Team and airport- and project-specific data entered by the user. ⢠Include the ability to review inputs. ⢠Allow flexibility in reporting and exporting the output of the cost estimating model, to support distribution of the results and exporting to other applications. ⢠Provide a what-if cost modeling capability. Following these guidelines, the Airport Capital Cost Estimation (ACCE) tool was developed in Microsoft Excel. The main interface of ACCE, shown in Figure 1, provides the interface to enter input required by the user, review interim results, and generate the cost estimating report. 4
Figure 1: ACCE main user interface Validating and refining the model Validation is a critical component of evaluating the performance and robustness of the cost model. The validation is an iterative process in which initial results are used to refine the model. Final validation serves to evaluate the quality of the model and identify needs for future work. Several validation techniques were used in this project: ⢠Statistical measures: The linear regression analysis used to develop the cost model results in the calculation of a number of metrics that can be used to evaluate the quality of the model and identify weaknesses. Key measures include the coefficient of determination (R2 and adjusted R2, the t-statistic for each coefficient in the model, and the F-statistic. 5
⢠Case study validation: In case study validation, an additional set of historical construction projects are collected for testing purposes. For each case study, the predicted cost per the model is compared against the actual cost. Conformance between predicted and actual cost across several case studies is a measure of the modelâs effectiveness and robustness. The case studies can also be analyzed in more details to better understand underlying factors that affect model performance. ⢠Subject Matter Expert (SME) feedback: The Research Team included two subcontracts that served as SMEs for horizontal and vertical construction projects respectively. The SMEs tested the model and evaluated it against their professional experience with cost estimating, to see how well the results matched expectations. 6
