39 Kwakkel Uncertainty In Airport Master Planning

AIRPORT RESEARCH

The Problem of Uncertainty in Airport Master Planning Airport strategic planning (ASP) focuses on the development of plans for the medium-term and long-term development of an airport. Strategic planning is defined as ‘the managerial activities that produce fundamental decisions and actions that shape and guide what the organization is, what it does, and why it does it’ (Bryson 1995: pp. 4-5, as cited in Bryson 2004). Strategic planning can be done in many different ways. In airports, the dominant approach is airport Master Planning (AMP), which results in a Master Plan that ‘presents the planner’s conception of the ultimate development of a specific airport’ (ICAO, 1987, pp. 1-2). In the US, the FAA has set up strict guidelines for an AMP study (FAA, 2005 and earlier versions). Internationally, IATA reference manuals as well as books about airport planning by leading scholars heavily influence AMP practices (e.g. ICAO, 1987; de Neufville and Odoni, 2003; IATA, 2004). By J.H. Kwakkel Uncertainty in Master Planning A crucial challenge in ASP is how to deal with uncertainty about the future. It is necessary to take into account the future world, which the organization will operate in, in order to make decisions that shape and guide what an organization is, what it does, and why it does it. In case of ASP, uncertainty is even more important, given the fact that decisions made today can shape and influence the airport performance for many years to come. For example, the decision to build a new runway at a specific location will likely influence the airport more than fifty years from now. It is therefore necessary to have a thorough assessment of potential developments that influence the future in

which the airport will operate, if one wants to plan effectively. In AMP, however, only demand uncertainties are considered, which are assessed through forecasting. Often, only a single demand forecast is created, and a Master Plan is designed based on that single forecast.

often has seriously negative consequences for the long-term development of an airport, including an inability to implement the plan, severe capacity constraints due to unanticipated (noise) regulations, an inability to meet aviation demand, and unnecessary investments in airside and landside facilities.

Criticism of Master Planning AMP and demand forecasting as the approach for the treatment of uncertainty in ASP has come under increasing criticism (see for example, de Neufville, 1991a; Walker, 2000; Flyvbjerg et al., 2003). The demand forecasts are practically always wrong, and Master Plans are often nearly impossible to implement. As such, AMP

Finding ways to deal with the many uncertainties surrounding the future of the air transport system is especially urgent in light of the fact that, in the coming years, the context in which airports operate is expected to become even more dynamic. Demand for air transport is expected to increase significantly, but there is uncertainty regarding numerous aspects, such as the extent of the

Picture 1: Artists impression of Berlin’s new BBI Airport. Courtesy Berlin Brandenburg Airport

e-zine edition, Issue 39

1

increase, on which routes, by which carriers, what the noise contours will be, what the regulatory regimes will be, etc. The United States, which was the first country to liberalize its air transport market, is already facing these issues. Hub operations can move from one airport to another, and airports are increasingly called upon to comply with the wishes of airlines (Dempsey et al., 1997; de Neufville and Odoni, 2003). The European Union started to liberalize its internal market in the mid 1990s, and this resulted in dramatic changes in air traffic demand. In addition, airports and airlines are being privatized, which introduces new stakeholders (e.g. within the financial market) and new requirements for airport performance. These developments lead to uncertainties about the future performance of the airport (e.g. capacity, delays, noise and financial performance), uncertainties about the robustness of the policies airport decision-makers want to implement, and uncertainties regarding the implementability of these policies. Together, these uncertainties hamper ASP and make the traditional AMP approach even less appropriate. The Challenge of uncertainty in ASP A general definition of uncertainty is ‘any departure from the unachievable ideal of complete determinism’ (Walker et al., 2003). Uncertainty is not simply a lack of knowledge, since an increase in knowledge might lead to an increase of knowledge about things we do not know and thus increase uncertainty ( ). The traditional way to deal with uncertainty in airport planning is through AMP. Based upon a limited number of forecasts of future traffic demand, a static plan is designed that can accommodate the forecasted traffic demand in an adequate way. It is assumed that the airport authorities are able to independently implement the plan without any opposition from other stakeholders (Dempsey et al., 1997; Burghouwt and Huys, 2003). Master Planning Airport Master Planning (AMP) is the process of developing a Master Plan. According to ICAO, the United Nations body for civil aviation, ‘an airport Master Plan presents the planner’s conception of the ultimate development of a specific airport’ (ICAO, 1987, pp. 1-2). This definition is also used by the International Air Transport Association (IATA) (IATA, 2004). According to the FAA, the United States regulator of aviation, ‘an airport Master Plan is a comprehensive study of the airport and typically describes short-, medium-, and e-zine edition, Issue 39

long-term plans for airport development’ (FAA, 2005)’, which is almost identical to the ICAO definition. The goal of a Master Plan is to provide a blueprint that will determine future airport developments (Dempsey et al., 1997; Burghouwt and Huys, 2003). As such, it describes the strategy of an airport operator for the coming years, without specifying operational concepts or management issues. A typical Master Plan, according to the FAA, should contain (i) a technical report containing the analyses conducted during the development of the Master Plan; (ii) a summary report that brings together facts, conclusions and recommendations relevant to a wider public; (iii) an airport layout plan drawing set which contains a graphical representation of the proposed developments in the Master Plan; and (iv) a website and public information kit for providing information about the Master Plan to the public (FAA, 2005). The time horizon covered in a Master Plan can vary, depending on the situation of the airport for which a Master Plan is developed. A short-term Master Plan has a time horizon of roughly five years, a mid-term Master Plan has a time horizon of six to ten years, and a long-term Master Plan has a time horizon of 20 years (FAA, 2005). AMP follows a strict linear process. The most commonly used guidelines (e.g. FAA, 2005; ICAO, 1987; IATA, 2004) are fundamentally the same, although they differ in detail (de Neufville and Odoni, 2003). The key steps in an AMP process are: > Analyze existing conditions;

> Make an aviation forecast; > Determine facility requirements; > Develop and evaluate several alternatives; > Develop the best alternative into a detailed Master Plan. Forecasting The aviation forecast is the main premise for a new Master Plan. By comparing the forecast with the existing conditions, an assessment can be made whether there is a need for new or expanded facilities. As such, aviation forecasting is the main way in which uncertainties about the future context, in which an airport operates, are handled. A forecast is a statement, usually in probabilistic terms, about the future state or properties of a system based on a known past and present. The basic concept of forecasting is simple: past trends, based on time series or theories about underlying mechanisms, are identified and extrapolated forward. In mathematical terms, a relationship between independent variables (X1, X2, …, Xn) and the dependent variable (Y) is developed: Y = f(X1, X2, …, Xn) that correlates well with past performance. This formula is then extrapolated to obtain a forecast for the year of interest. According to the FAA, forecasts should be realistic, based on the latest available data, reflect the current conditions at the airport, supported by information in the study, and provide an adequate justification for the airport planning and development (FAA, 2001). How are forecasts made? The first step in making forecasts is to identify the depend-

Picture 2:Berlin Tempelhof will be closed-down in favour for the new BBI Airport

2

ent variables (the Y’s) to forecast. Depending on the relevant issues and potential problems of a particular airport, the variables to estimate through forecasting generally include aircraft operations, passengers, air carrier enplanements, passenger enplanements, air carrier and commuter operations, tons of cargo, and aircraft operations by type of aircraft. The next step is to gather and analyze the data on the related independent variables (the X’s), which are assumed to be necessary to forecast the variables of interest. Relevant data sources include previous forecasts for the airport, historical aviation data, forecasts of other airports in the region, forecasts for air travel in the region, and socio-economic data. These data should be analyzed to see whether they are appropriate to be used and not ‘contaminated’ by unique events, such as a major sport event that created a temporary major boost in air traffic. The final step is to select a forecast method, such as regression and trend analysis or share analysis, and apply it in order to obtain an aviation forecast of Y for some future year. Forecasting is based on the idea of identifying trends and underlying mechanisms, based on the past and the present, and extrapolating them forward. However, it might be that the phenomenon you want to forecast has recently gone through changes, or is expected to undergo changes (e.g. trend breaks). In such situations, it is unwise to simply extrapolate based on past trends and known underlying mechanisms. In such situations, forecasters sometimes use trend break scenarios to produce forecasts that deviate from past trends (de Neufville and Odoni, 2003). The inadequacy of airport Master Planning AMP, as the main way to treat uncertainty, has proven to be ineffective. There are many examples of Master Plans that turned out to fail in practice (e,g, Nelkin, 1974; Nelkin, 1975; Szyliowics and Goetz, 1995; Demsey et al., 1997; de Neufville and Odoni 2003; Cidell, 2004). One example will be discussed here. An illustration: Amsterdam Airport Schiphol In 1995, after a two year process, which is known as the so-called physical planning key decision Schiphol (PKB Schiphol), a number of major decisions regarding the future of Schiphol Airport were made. The main objective was to facilitate the process of Schiphol becoming a mainport, while at e-zine edition, Issue 39

the same time improve the quality of living in the area surrounding the airport. Improving the quality of living should be measured in terms of noise, emissions, and third-party risk compared to the 1990 situation. Until 2003, only noise would be considered. Emissions and third-party risk would become relevant after that. The planning period was 20 years, from 1995 till 2015. Forecasts were created based on three scenarios in order to come to a decision. It became clear during the development of the PKB that in only one of the three scenarios both objectives could be achieved. The final PKB was based exclusively on the aviation forecasts derived from this scenario. The key decisions of the PKB were (Dutch Parliament, 1998-1999): > Schiphol would be allowed to grow into a small hub airport, with KLM as its hub carrier that should serve roughly 40 to 45 million passengers in 2015; > A fifth runway would be built; > Until 2003, which was the expected year the fifth runway would open, the noise situation should not get worse than the situation in 1990, which implies a maximum of around 15,000 houses in the high noise contour (the so called “stand still” principle); > After 2003, the maximum number of houses in the high noise contour would be lowered to 10,000 houses; > An insulation program for houses would be implemented within the high noise contour; > A study into the development of Lelystad airport to relieve Schiphol; > A high-speed rail link from the Netherlands to Belgium and France, and from the Netherlands to Germany, would be developed that would pass Schiphol, in order to reduce the number of short-haul flights from Schiphol. As it turned out, the limits of the noise regulations were reached in 1999, leading to a temporary shutdown of the airport, and the passenger limit was reached in 2005. The two-year costly process that aimed at developing a plan adequate for 20 years turned out to have a lifetime of less than ten years. How did this happen? The model used to create the demand forecast was based upon a relationship between GDP and traffic demand that represented past experience quite well. However, due to a number of trend breaks that happened after the forecasts had been made; this relationship no longer produced good predications, resulting in forecasts significantly lower than the

traffic demand actually experienced. The unexpected high rate of growth of air traffic demand was due to (i) an unanticipated rapid growth of the hub network of KLM, leading to an increase in transfer passengers; (ii) an alliance between KLM and Northwest Airlines, which fed passengers from both airlines through Schiphol; and (iii) The European Union’s liberalization process of the air transport industry, which increased competition among air carriers, lowering air fares, and paving the way for low-cost carriers. The inadequacy of aviation forecasting What is clear from the foregoing illustration is that AMP does not succeed in reaching its goals. Plans are quickly obsolete and are not robust with regard to the future. In other words, uncertainty (e.g. aviation demand, regulatory context, technological breakthroughs, and stakeholder behavior) is a key source of problems in ASP. One reason that AMP does not achieve its goals is that the only uncertainties that are considered are demand uncertainties, which are addressed through forecasting. However, forecasting has come under increasing scrutiny. Criticism can be split into two categories: forecasting failure due to bias, and forecasting failure due to uncertainty. Forecasters’ bias contributes to forecast failure in several ways. > Forecasters have a tendency to misjudge the relevance of (recent) data (Porter et al., 1991); > Forecasters often have a poor database that has internal biases caused by the data collection system (Flyvbjerg et al., 2003); > Forecasters often integrate political wishes into their forecasts (Flyvbjerg et al., 2003); > Forecasters use data from their home countries (instead of the local areas) for calibrating their models (Flyvbjerg et al., 2003); > Forecasts by project promoters may be even more biased, since the promoter has an interest in presenting the project in as favorable a light as possible (Flyvbjerg et al., 2003). Forecasting failure due to uncertainty manifests itself in several ways. As pointed out by Flyvbjerg et al. (2003), discontinuous behavior of the phenomena we try to forecast, unexpected changes in exogenous factors, unexpected political activities, and missing realization of complementary policies are important reasons for forecasting 3

failure. Ascher (1978) sees faulty core assumptions as a prime reason for forecasting failure. It refers to the fact that since the phenomenon we are trying to forecast is not completely understood, forecasters have to make assumptions about the data they need, the formula to be used, etc. (Porter et al. 1991). The use of historical data as a means of testing the adequacy of a given formula does not solve this problem, for there are an infinite number of formulas possible that can match the given historical data. Related to this is the fact that, in order to forecast a dependent variable Y based on a formula Y = f(X1, X2, …, Xn), we need forecasts for the future values of the n independent variables. Instead of forecasting a single variable, we end up forecasting n variables. Even if we were able to address the problems identified under the label of forecaster bias, this category of forecasting failure, due to uncertainty, implies that forecasting always can go wrong. By looking at the past and assuming that past behavior will continue in the future, we overlook a large part of the uncertainty that, when it manifests itself, will lead to trend breaks. Closing Remarks In conclusion, until now, a static reactive approach, in the form of Master Planning, to Airport Strategic Planning has dominated the airport planning and design process. In traditional AMP, the uncertainties are often ignored, oversimplified, handled probabilistically, or handled through the use of forecasts and scenarios. These methods have

proven insufficient for handling the uncertainties airports face, since they assume that the future is known to a degree that is sufficient for making appropriate decisions. Hence, finding new ways to deal with the different uncertainties surrounding the future is a key issue in air transport research. References Ascher W. 1978. Forecasting: an Appraisal for Policy-makers and Planners, Baltimore: Johns Hopkins University Press. Bryson J.M. 2004. “What to do When Stakeholders Matter: Stakeholder Identification and Analysis Techniques”, Public Management Review, Vol. 6, No. 1, pp. 21-53. Burghouwt G., Huys M., 2003. “Deregulation and the Consequences for Airport Planning in Europe”, DISP, 154, pp. 37-44. Cidell J.L. 2004. Scales of Airport Expansion: Globalization, Regionalization, and Local Land Use, July, 2004. Dempsey P.S., Goetz A.R., Szyliowicz J.S., 1997. Denver International Airport: Lessons Learned. McGraw-Hill, New York. Dutch Parliament, Tweede Kamer 1998-1999. Groeicijfers Schiphol; Rapport. 26265 nr. 2. obtained from http://www.rekenkamer.nl/9282000/d/tk26265_ 2.pdf on July 30 2007. Federal Aviation Administration (FAA). 2001. Forecasting Aviation Activity by Airport. . Washington D.C.: U.S. Department of Transportation. Federal Aviation Administration (FAA). 2005. Advisory Circular 150/5070-6B, Airport Master Plans. Washington D.C.: U.S. Department of Transportation. Flyvbjerg B., Bruzelius N., Rothengatter W. 2003. Megaproject and Risk: an Anatomy of Ambition,

Cambridge, Cambridge University Press. IATA, International Air Transport Association 2004. Airport Development Reference Manual. Montreal, Canada. ICAO, Interational Civil Aviation Organization 1987. Airport Planning Manual, Part 1, Master Planning, Montreal Canada. Nelkin D. 1974. Jetport: the Boston Airport Controversy, New Jersey, Transaction Books. Nelkin D. 1975. “The Political Impact of Technical Expertise”, Social Studies of Science, Vol. 5, No. 1., pp. 35-54. de Neufville, R. and Odoni, A. 2003. Airport Systems: Planning, Design, and Management. . New York: McGraw-Hill Professional. Porter A.L., Roper A.T., Mason T.W., Rossini F.A., Banks J. 1991, Forecasting and Management of Technology, John Wiley & Sons, New York. Szyliowicz J.S., Goetz A. R. 1995, “Getting Realistic about Megaproject Planning: The Case of the New Denver International Airport”, Policy Sciences, Vol. 28, No. 4, pp. 347-367. Walker W.E., Harremoës P. Rotmans J., Sluis J.P. van der, Asselt M.B.A. van, Janssen P., Krayer von Kraus M.P. 2003. “Defining Uncertainty: A Conceptual Basis for Uncertainty Management in Model-Based Decision Support”, Integrated Assessment, Vol. 4, No. 1, pp. 5-17. Walker, W.E. 2000. ‘Policy Analysis: A Systematic Approach to Supporting Policymaking in the Public Sector’, Journal of Multicriteria Decision Analysis, Vol. 9, Issue 1-3, pp. 11-27.

About the Author J.H. Kwakkel is PhD researcher at the Faculty of Technology, Policy and Management of the Delft University of Technology. [email protected]

Picture 3: Computer animation of BBI’s new terminal. Courtesy Berlin Brandenburg Airport.

e-zine edition, Issue 39

4