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 4: CONCLUSIONS AND SUGGESTED RESEARCH After an iterative process consisting of several rounds of data collection, the Research Team created a cost database consisting of 251 observations of historical airport construction projects. The parametric cost estimating methodology was used to develop CERs for eight construction types. The CERs were developed using linear regression analysis, using ordinary least squares estimation. A user interface developed in Microsoft Excel was employed to provide a simple but effective mechanism for members of the airport community to interact with the cost model. The cost model was tested and validated through a series of statistical test, by visual inspection of plots of predicted versus actual cost, and through subjective evaluation by the Research Teamâs airport construction SMEs. The statistical measures used to test the quality of the model were satisfactory, both at the individual CIV level and when considering the CER as a whole. The plots of predicted versus actual cost show general adherence to the expected reference line, but scatter about the line indicate that the performance of the CERs is variable. The case study validation also confirms that the model can be highly variable. For this reason, the accompanying Guidebook includes considerable guidance on how the results should be interpreted. The Guidebook identifies the limitations of the model, including checklists to identify specific sources of uncertainty. The variable performance of the model is a result of the limited size of the database, which in turn is a function of limited data availability. While the stakeholder outreach effort indicates that a large number of airports and agencies store historical cost data, there are a number of challenges that limit the value of the available information. These challenges include: ⢠When historical cost data is available it is often stored in a format that prevents from easily being incorporated into an electronic spreadsheet or database. For example, data is often stored in PDF format, which is an electronic format, but which cannot readily be imported into a database. ⢠Airport projects often bundle several construction activities that represent different project types. This can make it difficult or impossible to attribute construction cost to a specific project type, which is necessary in order to support CER development. ⢠Historical project data often does not include information on the CIV values required to develop CERs. For example, information on a pavement project may include the amount of pavement materials required, but not specify the pavement surface area or the design aircraft MTOW. ⢠When airport projects are funded with AIP or PFC funding, the available cost data may only specify the federal share. This can make it difficult to identify the total project cost, especially if the project includes a large share of ineligible items. 23
Overall, however, the Research Team concludes that the objective of producing a cost database and model based on parametric cost estimating has been met. Recommendations for future work focus on establishing guidelines for expanding the cost database. Expanding the database would allow for additional project types to be supported in the cost model. It would also allow for a larger number of CIVs to be included in the CERs. This should improve the explanatory power of the CERs, leading to increased robustness and substantially less variability in the modelâs performance. The Guidebook prepared by the Research Team contains specific recommendations for data collection practices, including guidance on defining total project cost and data requirements that would support an increase in the number of cost drivers that can be included in the model. 24
