Handbook to Assess the Impacts of Constrained Parking at Airports (2010)

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.

49 A menu of strategies airport operators can implement to resolve ongoing public parking constraints or manage con- strained parking events was provided in Chapter 5. Implemen- tation of many of those strategies requires advanced planning and consideration of the impacts the strategy may have on related issues, such as parking facility use, capital or operational costs, parking-derived revenue, vehicle traffic and emissions, and customer service. Implementation of strategies intended to resolve an ongoing public parking constraint may require capital investment and possibly executive or regulatory approval, whereas implemen- tation of strategies to address shorter-term constrained parking events tends to involve a lower level of effort and investment. For those strategies that require more time and investment to implement, decision makers need to understand whether the strategies are likely to achieve the desired outcomes, especially within the framework of an airport operator’s goals and objec- tives for its parking program. The formal and informal tools and methodologies for pre- dicting the outcome of strategies implemented to resolve ongo- ing constrained public parking are described in this chapter. Some of these tools and methodologies may also be useful to predict the outcome of strategies to manage short-term parking constraints. This research project included the development and evalu- ation of the usefulness of a formal predictive tool (i.e., a model) to assist airport operators in understanding the magnitude of changes in parking behavior resulting from implementation of a strategy. Such a model could reveal unanticipated outcomes that would influence decision making. The consequences of a lack of understanding of the potential outcomes in advance of strategy adoption and implementation can be severe. Formal Tools Formal tools can be used to predict the outcomes of strate- gies being considered to address constrained parking. Three formal tools are discussed in this section—airport mode choice models, airport parking models, and an airport parking fore- cast model developed based on the research conducted for ACRP Project 10-06. Airport Mode Choice Models A regional organization, such as an MPO, is often respon- sible for regional travel demand forecasting, and may main- tain a regional travel demand forecasting model. In some cases, modeling efforts my be initiated by or coordinated with an airport operator to incorporate airline passenger O&D sur- vey data into the regional travel demand forecasting model. This airport mode choice module can provide estimates of changes in mode share and trip volumes based on various parking and transportation policy changes. Such a modeling effort requires significant input from the airport operator as the entity most familiar with the ground access travel patterns of airport customers. Of the 15 airport operators participating in this research project, 2 have developed this type of predictive tool—the Massachusetts Port Authority for BOS and the Port of Port- land for PDX. These two examples are discussed in the follow- ing sections. Logan Mode Choice Model The Boston Region MPO, Central Transportation Planning Staff, maintains and operates a regional travel demand model that encompasses the 101 cities and towns that form the Boston metropolitan area. This model has a nested airport mode choice model, referred to as the Logan Mode Choice Model, which is used to estimate trips generated by BOS air- line passengers. This model is calibrated to airline passenger travel behavior using airline passenger O&D survey data col- lected every 3 years by Massport, owner and operator of BOS. The most recent calibration was completed in 2007. The Logan Mode Choice Model is a multinomial nested logit model used to estimate access trips to BOS based on C H A P T E R 6 Predicting Outcomes of Selected Strategies

50 15 travel modes, consistent with the categories in the airline passenger O&D survey. These travel modes include detailed information on private automobile use—use of the curbside only for pickup and drop-off, use of short-term parking for pickup and drop-off, or use of parking for the duration of the airline passenger’s trip. The model produces an estimate of trips for the four main airline passenger customer segments— resident business, resident nonbusiness, nonresident business, and nonresident nonbusiness. It incorporates trip character- istics that influence airline passenger airport access choices, including trip duration, travel party size, number of bags checked, whether an employer is paying travel costs for a busi- ness trip, and other factors that affect mode choice decisions. The model can be used to (1) constrain BOS public parking demand to the existing number of spaces; (2) estimate trips by mode at different parking rates; (3) estimate trips for different fares on HOV access modes; and (4) estimate the effects of changes in airport transit service, regional transit service, and the regional highway system. This information is then used to calculate estimated changes in VMT by airline-related passen- ger trips on local and regional roadways. Airport employee trips are not included in the Logan Mode Choice Model. Trips generated by airport employees are esti- mated in the regional model; however, the regional travel demand model does not account for the unique trip patterns of airport employees. (1) Airport Passenger Demand Model In fall 2009, the Port of Portland, owner and operator of PDX, finalized an update to the airline passenger ground access travel component of the regional travel demand model referred to as the Airport Passenger Demand Model (APDM). The MPO maintains the regional travel demand model. The APDM was developed to provide the following: • A spreadsheet model that enables the Port of Portland to test scenarios related to changes in airline passenger ground access costs, travel times, and transit availability to determine the change in trip distribution and mode choice to PDX. • An application code that is tied to the regional travel demand model and is used to estimate the outcome of certain strate- gies related to parking, expansion of the regional transit sys- tem, and other measures at the regional level. The APDM is a multinomial nested logit model that esti- mates trips for four airline passenger customer segments: resi- dent business, resident personal, visitor business, and visitor personal. Trips are distributed among 11 modes, which include long-term parking in each of the three on-airport parking facil- ities as well as privately operated off-airport parking, and pickup or drop-off in private automobile with and without the use of short-term parking. The model was developed using approxi- mately 2,000 responses from passenger intercept surveys col- lected from O&D airline passengers at PDX traveling during June and September 2008. Passenger intercept survey data indi- cate that resident business travelers value their travel time twice as much as residents and visitors whose travel purpose is per- sonal. Visitors traveling for business valued their time the most—2.5 times more than resident business travelers. Airport employee trips are not included in APDM. Trips generated by airport employees are estimated in the regional model, which does not account for the unique trip patterns of airport employees. The Port of Portland has tested multiple scenarios with its spreadsheet model to determine how airline passenger mode choices would shift with increases in parking rates, changes in costs of other travel modes, changes in travel times, and changes in the frequency of transit services. The results, shown in Table 6, are illustrative of how the model can predict changes to assist the Port in determining the outcome of policy changes Table 6. Portland International Airport airline passenger mode share with application of different transportation policy scenarios. Policy Scenario Mode Shares Parking Charge Increase Access Mode Existing Mode Share 10% 25% 200% Double Travel Time Increase Travel Time and Automobile Operating Costs by 20%, Provide Free Transit at Double the Frequency, and Double the Parking Costs Drive and Park 34% 31% 27% 10% 37% 16% Pickup and Drop-off 33% 36% 39% 54% 25% 44% Taxicab, Limousine, Town Car 6% 6% 6% 7% 6% 6% Rental Car 17% 17% 17% 17% 17% 17% Shuttle 4% 4% 4% 4% 4% 4% Transit 6% 6% 7% 8% 11% 13% Total 100% 100% 100% 100% 100% 100% Source: Port of Portland, October 2009. (19)

51 it may introduce, and how factors and policies external to its control or authority, such as vehicle traffic congestion and transit service frequency provided to PDX by TriMet (the regional transit authority) influences airline passenger choices. To summarize, the model results indicate that increased parking rates primarily result in shifts to pickup and drop-off modes, which double vehicle trips to PDX compared to park- ing for the duration of the airline passenger’s trip. Increasing parking rates would reduce the demand for parking, but it would also increase vehicle traffic congestion and emissions generated by airline passenger access and egress trips. Even a tripling of parking rates would only increase the transit mode share from 6% to 8%. Increasing transit service and offering transit service free of charge, options that are not within the control of the Port of Portland, would only increase the tran- sit mode share from 6% to 7% (this scenario is not shown in the table, but it was tested by the Port of Portland). Therefore, improving transit options as a stand-alone strategy to relieve constrained parking would not appear to be a successful strat- egy for managing constrained parking. (19) Airport Parking Models An airport operator also may develop its own predictive tool (i.e., a model) for parking demand or for all airline pas- senger modes. The model may be developed as part of the parking management program, the airport ground access pro- gram, or as part of a larger project, such as a master plan. The Port of Seattle developed such a model to predict parking behavior under a variety of scenarios that consider pricing, capacity, and other factors. Seattle-Tacoma International Airport’s Composite Parking Demand Model In late 2008, the Port of Seattle initiated an effort to gain a better understanding of the underlying causes of fluctuations in parking demand at Seattle-Tacoma International Airport (SEA). Various multiple regression econometric models were constructed to help forecast enplanements and parking trans- actions in the main garage at SEA. The econometric models included independent variables, such as per capita income, population, airfare, and employment, while also taking into account seasonal variations in demand and major shifts in demand due to one-time events, such as the terrorist attacks on September 11, 2001. The effort concluded with the development of various mod- els to help forecast and strategize garage parking pricing, park- ing demand and capacity requirements, and new program implementation. By using the models, the Port of Seattle has been able to gain a better understanding of the relationship between key economic conditions and the demand for airport garage parking. Prior to development of the models discussed here, the Port of Seattle had developed a parking choice model with price elasticity curves to analyze the parking market shares of its one public parking facility, the main garage, versus off-airport lots in the vicinity of SEA. In 2008, several analytical model steps that led into, and interfaced with, the parking choice model were added. The sequence and components of the expanded model set, referred to as the Composite Parking Demand Model, are presented below. 1. Enplanement Model—The Enplanement Model and its two submodels for originating and returning passengers were developed to provide an analytical tool for forecasting future volumes of these two specific groups of passengers that form the basis for demand for ground transportation access to SEA. 2. Main Garage Demand Model—The Main Garage Demand Model and its three submodels used to predict parking demand for weekly, daily, and hourly parking transactions were developed to provide a tool for forecasting the future parking activity of three distinct groups of parking cus- tomers driven by different sets of economic and travel con- siderations. This set of three submodels was also designed to support estimation of the capacity and revenue impacts in the main garage of different pricing scenarios and of fluc- tuations in the local economy. These submodels add to the Port of Seattle’s analytical toolkit for evaluating policy options related to the main garage. Combining enplanement forecasts from the Enplanement Model with forecasts of the various factors included in the main garage submodels sup- ports both parking transaction forecasting efforts and sensi- tivity analysis of the effects of future scenarios involving the sets of influential variables in the three submodel equations. 3. Monthly Transactions and Duration Conversion Model— The Monthly Transactions and Duration Conversion Model is a submodel of the Composite Parking Demand Model. Its purpose is to translate the parking transactions forecast information from the three main garage submodels into a detailed format conformable to the Parking Choice Model. 4. Parking Choice Model—This model was developed to ana- lyze data on the main garage versus off-airport parking lot shares of overall airline passengers parking in the Seattle- Tacoma area. It includes detail on the main garage shares of the market for parking transactions of different durations, provides information on the main garage versus off-airport prices for parking transactions of different durations, and derives price response coefficients based on how much the main garage share declines for longer stays as its price pre- mium over off-airport options becomes greater. Originally calibrated for 2006–2007 transactions, the Parking Choice Model is structured so that alternative transaction data, such as that produced by the Monthly Transactions and Duration

52 Conversion Model, can be input. By explicitly including variables for off-airport parking prices, the model also allows the user to examine the composite effects of alterna- tive pricing scenarios, such as main garage pricing changes that are mirrored by off-airport lot pricing reactions, or main garage pricing changes alone. (20) ACRP Project 10-06 Airport Parking Forecast Model A model for testing resident airline passenger mode-share behavior was developed for ACRP Project 10-06 to provide analysts with a tool for predicting, at a high level, likely out- comes of strategies being considered to address constrained airport parking. The model, referred to as the General Airport Parking Forecast Model, was developed based on data col- lected at 14 U.S. airports. An airport-specific version of the model was also developed based on data collected at PDX, which was among the 14 airports used for data collection. These two versions of the model were developed to provide airport operators and others with information regarding the benefits and tradeoffs of developing their own airport-specific models versus using the general airport model. The research team evaluated the effectiveness of the models by comparing projected results of identical policy scenarios for PDX from the airport-specific model and the general airport model, and found both to be effective for estimating the results of policy scenarios, as described in the rest of this section. The general airport model was developed as a tool for any commercial airport operator to use to estimate results at its airport without undergoing an extensive data collection effort. The model can be used to compare results between scenarios for a specific airport. It calculates the results using the under- lying survey data from the 14 U.S. airports and the character- istics of the available modes and resident airline passenger mode shares for the specific airport that would be added to the inputs page by the analyst. The general airport model and instructions on how to use it are available on the CD-ROM that accompanies this report. In comparison to the general airport model, an airport- specific model will produce estimated results with a higher level of accuracy for the airport, may include more mode options that are specific to the airport, and can be structured to test more strategies and strategies that are more relevant to the specific airport environment. Such a model requires an invest- ment of time and money by the airport operator, including the collection of survey information for model development. The airports included in this research have experienced constrained parking conditions within the past 10 years. The airport-specific model is based on data collected at PDX. The general airport model is based on data collected at the follow- ing airports: • Boston Logan International Airport (BOS), • Chicago O’Hare International Airport (ORD), • Huntsville International Airport (HSV), • McCarran International Airport (LAS), • Miami International Airport (MIA), • Oakland International Airport (OAK), • Port Columbus International Airport (CMH), • Portland International Airport (PDX), • San Antonio International Airport (SAT), • San Diego International Airport (SAN), • Seattle-Tacoma International Airport (SEA), • Tampa International Airport (TPA), • Tulsa International Airport (TUL), and • Washington Dulles International Airport (IAD). To provide reliable results for the airport-specific model, the sample size of the data collected at PDX was larger than the sample sizes collected at the other 13 airports for the General Airport Parking Forecast Model. The PDX sample was weighted for inclusion in the general airport model so it would not skew the results of the general model. The models were developed in Microsoft Excel and offer user-friendly interfaces. They are multinomial logit models based on data collected in an online stated preference survey of airline passengers at each of the sample airports, conducted between April 21, 2009 and May 4, 2009. Stated preference survey data are useful in estimating cause–effect relationships for airport access. A stated preference survey is designed to collect much of the information obtained in an O&D survey that is necessary to understand the respondent’s ground access behavior on a previous trip (referred to as “revealed prefer- ence” data), including mode, trip purpose, trip origin, travel party size, length of stay, and other relevant information, as well as data on future ground access choices airline passengers would make under different policy scenarios, referred to as “stated preference” experiments. Stated preference experi- ments were used to test the effects on airport choice and parking behavior of a wide range of variables that are likely to influence decisions on whether or not to use airport parking, including location, price, availability, shuttle service quality and availability, and availability and level of service of alterna- tive HOV options to access the airport. An example of a stated preference experiment from the survey is shown in Figure 2. The methodology used for collecting the stated preference sur- vey data, as well as the survey instrument, are provided in the Final Report for ACRP Project 10-06. A discussion of both the general airport and airport-specific parking forecast models is provided in the following sections. The level of reliability and results of the general airport model also are discussed. To achieve more detailed results, airport operators may consider developing a model specific to their airport conditions and characteristics, so a comparison of the

53 two models and a discussion of approaches for estimating the effects of strategy implementation from the model results also are provided. General Airport Parking Forecast Model The General Airport Parking Forecast Model captures the difference between large-hub airports and small- or medium- hub airports and can be used to test strategies at any small-, medium-, or large-hub airport. The model provides planning- level insight into potential airport operator and other trans- portation agency policies to address constrained parking. As such, it is a useful tool for airport policymakers to use in eval- uating a range of potential strategies being considered by reviewing the relative changes in mode shares for each strat- egy tested. Examples of strategies that may be tested with the model include changes in parking rates, changes in the level of service of remote parking shuttles, changes in the level of service or fares for HOV modes, the introduction of transit at airports that do not offer transit, the addition of remote parking capac- ity, and a drop-off fee for private automobiles transporting passengers in the pickup and drop-off mode. The model does not allow the user to test the relationship between the use of parking owned by the airport operator and privately operated parking. Privately operated parking is included as part of the mode category for park and ride shuttle to terminal. The model also does not have the capability to account for the severity of the parking constraint at an airport. The Final Report for ACRP Project 10-06 provides recommendations for future enhancements to the model to address these limitations. The research team used industry standard modeling meth- ods to determine model segments and coefficients that best fit the stated preference data set. During model estimation, it was observed that behavioral differences existed between business and nonbusiness trips that could be captured by using separate choice models. The two separate choice models were incorpo- rated into the General Airport Parking Forecast Model. From this, the Excel-based forecast model was created by calculating the probability of using an access mode for a specific scenario and by applying the probability to the sample to calculate respondent-level preferences for each access mode. Source: Resource Systems Group, Inc., 2009. Figure 2. Example of a stated preference experiment from ACRP Project 10-06 Stated Preference Survey.

54 Using the General Airport Parking Forecast Model. The general airport model, created in Microsoft Excel 2007, is avail- able on the CD-ROM that accompanies this report. Model inputs include base case resident airline passenger mode-share distribution data for resident airline passengers traveling for business and nonbusiness purposes, and travel times and rates for parking and other modes, as shown in Table 7. The travel time pricing inputs are also shown in Table 8. The most accu- rate source for resident airline passenger mode-share data is a survey of O&D airline passengers, as described in Chapter 8. Step-by-step instructions for testing strategies for alleviat- ing constrained public parking in the general airport model are as follows: 1. Enter base case mode share for resident business and res- ident nonbusiness passengers into base case ground access mode shares cells. If the mode shares for resident business and resident nonbusiness passengers are unknown, the mode share for resident airline passengers can be entered into both sets of cells. As noted previously, the model does not distinguish between privately operated off-airport park- ing and remote public parking. Both categories are included in “park and ride shuttle to terminal.” 2. Enter the proportion of business and nonbusiness resident airline passengers in the base case column. 3. Enter actual or estimated base data for pricing and travel times into the base case column. 4. Enter pricing and travel times appropriate for the strategy being considered. 5. Click the cursor on “calibrate to base case” button. Instruc- tions for how to set macro permissions in Excel for optimum use of the model are included in model documentation. 6. The mode-share distribution will be shown in the model output section of the user interface page. Table 9 presents an example of results from the model output tables. Application of the General Airport Model. The Gen- eral Airport Parking Forecast Model can be used to compare the relative effects of many different strategies being con- sidered. Strategies may be tested individually or together. To understand the effect of each strategy, the strategies should be evaluated individually before considering the adoption and implementation of a combination of strategies. Common examples of strategies an airport operator may consider include the following: • Parking rate changes at different parking facilities (the model allows the user to test parking rate changes for the parking supply within walking distance of the terminal and the remote parking supply, but does not distinguish park- ing beyond these two categories; rate changes may include absolute dollar increases for terminal area and remote park- ing, uniform percentage increases to each, increases in only one category, or changes that examine the relationship between pricing in each category); • Improvements or reductions in shuttle service between the terminal and remote parking facilities, which include the frequency to the terminal or wait times; • Improvement or degradation in the travel times of other modes in relation to automobile travel time; and • Changes in passenger fares for modes that are alternatives to driving and parking, including instituting a drop-off fee for automobiles transporting airline passengers for pickup and drop-off at the curbside. Three example policy scenarios were tested in the general airport model. The example policy scenarios, along with an interpretation of the model results, include the following: • General Airport Model Scenario No. 1: Doubling of Park- ing Fees—One of the key strategies an airport operator will consider to try to influence parking mode share is to change parking rates. In the scenario shown in Table 10, a dou- bling of the parking fees at a hypothetical small-hub airport is tested. Although airport operators do not frequently dou- ble the fees for public parking, the example illustrates how a dramatic change in parking fees would affect travel behav- ior. For purposes of this example, doubling parking fees is representative of constraining parking because this dramatic increase in parking fees is likely to influence passenger per- spective of the availability of parking. In this scenario, 16% of total airline passengers accessing the airport (account- ing for 40% of the passengers that would have parked for the duration of their trips) shifted primarily to the drop- off mode (11%), followed by the taxicab mode (2%). This analysis illustrates the relationship between parking con- straints and shifts to drop-off modes. • General Airport Model Scenario No. 2: Reduction of Parking Fees—A second scenario was tested to measure the influence that a 50% reduction of parking fees would have on passenger ground access behavior at a large-hub airport. This scenario, shown in Table 11, is presented to demonstrate the relationship between changes in perceived parking constraints (or, in this case, reduced constraints) and ground access mode-share distribution. In this sce- nario, the airport has a significant transit mode-share of 15%; however, this mode share was minimally affected by the policy to influence passenger parking behavior. The shift in mode share occurred from drop-off modes (private automobile and taxicab) to the use of parking facilities. • General Airport Model Scenario No. 3: Addition of Parking—The presence of off-airport parking has a mean- ingful effect on an airport’s ground access mode-share dis- tribution. Although the stated preference survey did not distinguish between on-airport parking that required a

55 Airport Specific Base Case & Policy Scenario Levels Base Case Policy Scenario Units Park & Walk to Terminal Parking Fee $25.00 $35.00 pe r da y Park & Ride Parking Shuttle to Terminal Parking Fee $18.00 $20.00 pe r da y Parking Shuttle Riding Time to Terminal 10 10 mi nu te s Wait Time for Shuttle 10 5 mi nu te s Airport Drop Off Charge N/A $0.00 $/t ri p Taxi/Limo/Towncar Fare by Distance $2.0 0 $2.50 $/ mi le Transit Fare $3.0 0 $3.50 $/t ri p Shared Van Fare by Distance $1.7 5 $2.00 $/ mi le Scheduled Bus Fare by Distance $0.2 0 $0.20 $/ mi le Additional Transit Time (over auto travel time) 0.30 0.30 mi ns /m il e Additional Shared Van Time (over auto travel time) 0.30 0.30 mi ns /m il e Additional Bus Time (over auto travel time) 0.30 0.30 mi ns /m il e Amount of Remote Parking 1.00 1.20 (1 , 000 s of s pac es ) Alternative Availability Base Case Policy Scenario Park & Walk to Terminal TRUE TRUE Park & Ride Shuttle to Terminal TRUE TRUE Taxi/Limo/Towncar to Terminal TRUE TRUE Dropped Off at Terminal TRUE TRUE Transit to Airport TRUE TRUE Shared Van to Airport TRUE TRUE Scheduled Bus to Airport TRUE TRUE Resident Air Passengers Trip Purpose Base Case Business Trips 29% Non-Business Trips 71% Airport Size Small/Medium Hub Base Case Ground Access Mode Shares Business Trips Park & Walk to Terminal 32% Park & Ride Shuttle to Terminal 17% Taxi/Limo/Towncar to Terminal 17% Dropped Off at Terminal 14% Transit to Airport 10% Shared Van to Airport 9% Scheduled Bus to Airport 1% Total 1 00% Base Case Ground Access Mode Shares Nonbusiness Trips Park & Walk to Terminal 13% Park & Ride Shuttle to Terminal 18% Taxi/Limo/Towncar to Terminal 9% Dropped Off at Terminal 32% Transit to Airport 18% Shared Van to Airport 7% Scheduled Bus to Airport 3% Total 1 00% Source: Resource Systems Group, Inc., November 2009. Table 7. Model inputs for the General Airport Parking Forecast Model.

Model Input Units Park and Walk to Terminal Parking Fee Per day Park and Ride Shuttle to Terminal Parking Fee Per day Shuttle Riding Time to Terminal Minutes Wait Time for Shuttle Minutes Airport Drop-Off Charge1 $ per trip Taxi Fare by Distance $ per mile Transit Fare $ per trip Van Fare by Distance $ per mile Scheduled Bus Fare by Distance $ per mile Additional Transit Time (over automobile travel time) Minutes per mile Additional Van Time (over automobile travel time) Minutes per mile Additional Bus Time (over automobile travel time) Minutes per mile Amount of Off-Airport Parking Spaces (in thousands) Note: 1 The drop-off fee is a per trip fee charged to all passengers dropped off by private automobile at the terminal. Source: Resource Systems Group, Inc., August 2009. Table 8. Pricing and travel time model inputs, General Airport Parking Forecast Model (base case). Resident Access Mode Share Base Case Policy Scenario Absolute Difference % Difference Park & Walk to Terminal 32% 23% -9% -30% Park & Ride Shuttle to Terminal 17% 23% 6% 35% Taxi/Limo/Towncar to Terminal 17% 14% -3% -17% Dropped Off at Terminal 14% 19% 5% 35% Transit to Airport 10% 11% 1% 14% Shared Van to Airport 9% 9% 0% -1% Scheduled Bus to Airport 1% 1% 0% 38% Total 100% 100% Resident Access Mode Share Base Case Policy Scenario Absolute Difference % Difference Park & Walk to Terminal 13% 9% -4% -35% Park & Ride Shuttle to Terminal 18% 19% 1% 5% Taxi/Limo/Towncar to Terminal 9% 7% -2% -21% Dropped Off at Terminal 32% 37% 5% 15% Transit to Airport 18% 19% 1% 4% Shared Van to Airport 7% 7% 0% -7% Scheduled Bus to Airport 3% 4% 1% 17% Total 100% 100% Resident Access Mode Share Base Case Policy Scenario Absolute Difference % Difference Park & Walk to Terminal 19% 13% -6% -32% Park & Ride Shuttle to Terminal 18% 20% 2% 13% Taxi/Limo/Towncar to Terminal 11% 9% -2% -20% Dropped Off at Terminal 27% 32% 5% 18% Transit to Airport 16% 17% 1% 6% Shared Van to Airport 8% 7% 0% -5% Scheduled Bus to Airport 2% 3% 0% 19% Total 100% 100% Business Trips Nonbusiness Trips All Trips Source: Resource Systems Group, Inc. 2009. Table 9. Example of output from the General Airport Parking Forecast Model.

57 shuttle bus to access the terminal and privately operated off-airport parking, a scenario was tested in the constrained parking forecast model in which 5,000 remote spaces (cor- related to the “park and ride shuttle to terminal” mode) were added to the public parking supply. This scenario applies to the addition of 5,000 spaces to either the on-airport public remote parking supply or the off-airport privately operated parking supply. Table 12 presents the results of this sce- nario. The addition of remote parking shifts mode share mainly from the “park and walk to terminal” and “dropped of at terminal” modes to the “park and ride shuttle to termi- nal” mode. This result implies that the addition of parking capacity does not generate significant demand for parking, as the overall share of passengers parking increased only two percentage points despite a significant increase in parking supply. In this circumstance, with the majority of the mode shift coming from the “park and walk to terminal” mode and the “dropped off at terminal” mode, vehicle trips to the airport and in the terminal area will decrease because, for every one-way airline passenger trip, passengers who are dropped off generate two vehicle trips and passengers who park for the duration of their trips only generate one vehicle trip. However, revenue implications to the airport operator also would need to be considered. As demonstrated in the example scenarios, the General Airport Parking Forecast Model can be used to test a variety of changes (price and travel time) related to the provision of airport access modes that could be used to address constrained parking conditions. The estimates from this model represent averages from the 14 airports surveyed and airport-to-airport differences may not be fully represented when the model is applied to a specific airport. The General Airport Parking Forecast Model reasonably represents the general magnitudes of changes in airline pas- senger access mode choices, even for those airports that do not All Trips Access Mode Share Base Case Policy Scenario1 Absolute Difference Percent Difference2 Park and Walk to Terminal 15% 6% -9% -57% Park and Ride Shuttle to Terminal 25% 18% -7% -27% Taxicab to Terminal 10% 12% +2% +22% Dropped Off at Terminal 40% 51% +11% +27% Transit to Airport 1% 1% 0% +21% Shared Van to Airport 4% 5% +1% +22% Scheduled Bus to Airport 5% 6% +1% +27% Total3 100% 100% Notes: 1 Although this scenario is representative of conditions at a small-hub airport, which is less likely to be well served by public transit compared to large-hub airports, the stated preference survey experiments did include public transit options. 2 Percent difference calculations may differ due to rounding. 3 Totals may not add to 100% due to rounding. Table 10. Doubling of parking fees at a small-hub airport. All Trips Access Mode Share Base Case Policy Scenario Absolute Difference Percent Difference1 Park and Walk to Terminal 5% 11% +6% +117% Park and Ride Shuttle to Terminal 10% 15% +5% +48% Taxicab to Terminal 30% 27% -3% -10% Dropped Off at Terminal 30% 24% -6% -19% Transit to Airport 15% 15% 0% -3% Shared Van to Airport 5% 4% -1% -11% Scheduled Bus to Airport 5% 4% -1% -19% Total2 100% 100% Notes: 1 Percent difference calculations may differ due to rounding. 2 Totals may not add to 100% due to rounding. Source: Resource Systems Group, Inc., August 2009. Table 11. Reduction of parking fees by 50 percent at a large-hub airport.

58 have highly accurate access mode-share information available, and should prove useful as a planning-level model for any airport with constrained parking. When the General Airport Parking Forecast Model is used and calibrated to the base access mode shares for the airport under study, the results should be interpreted as accurately representing relative changes when comparing pricing and other policies. For exam- ple, the differences in the mode-share distribution that results from a 10% increase in terminal parking prices compared to those that result from a 20% increase in terminal parking prices should be accurately represented (e.g., within 15% or so, based on variations in behavior among airports as observed in the models developed for this research project), as well as the dif- ferences in mode-share distribution that result from the sce- narios presented in the example scenarios. However, the actual mode shares that result from pricing or policy changes may differ from the model-estimated shares because of differences in behavior among airports that are not represented in the general airport model. An airport’s specific characteristics could be represented in more detail and provide a higher level of predictive accuracy if an airport-specific survey and model were developed, as described in the next section. Airport-Specific Parking Forecast Model An airport-specific parking forecast model is a customized model that represents the environment of a specific airport, thereby increasing the model’s overall utility. In comparison to the General Airport Parking Forecast Model, a model devel- oped specifically for one airport may include more mode options that are specific to that airport and could be struc- tured to test more strategies and strategies that are more rele- vant to that specific airport environment. The airport-specific model may also include additional calculations of outcomes related to strategies that have the potential to resolve con- strained airport parking, such as gross parking-derived rev- enues and the likely changes in the number of vehicle trips generated by airline passengers. An airport-specific parking forecast model was developed for PDX as part of this research project. The model included access modes, pricing, and time variables specific to PDX. As part of the sampling plan, a larger number of responses was collected from the PDX catchment area than was collected in the sample from each airport for the General Airport Parking Forecast Model. This larger sample from a single airport allowed for the development of a model specific to circumstances at PDX. PDX was selected because (1) the Port of Portland has dealt with policy-related constrained parking conditions since 2003, (2) a light rail line to PDX opened in 2001, and (3) the public parking supply at PDX is supplemented by a privately operated off-airport parking supply. In addition, the Port of Portland was one of two airport operators participating in this research project that had the potential to field-test results and compare them to similar results from their own predictive tools. The model was developed using a similar methodology as described for the General Airport Parking Forecast Model, except that separate modules were not developed for business and nonbusiness passengers. The draft PDX parking model was field tested by a Port rep- resentative to obtain feedback on the model with respect to its ease of use and applicability of results. The model was received favorably, except for a preference to segment the mode-share distribution by business and nonbusiness travelers. It was noted that the results for strategies tested were similar to the results from the PDX Air Passenger Demand Model. Value of Airport-Specific Parking Forecast Model Development of an airport-specific parking forecast model will most likely require new data collection and model devel- opment by the airport operator or another interested party, All Trips Access Mode Share Base Case Policy Scenario Absolute Difference Percent Difference1 Park and Walk to Terminal 5% 4% -1% -25% Park and Ride Shuttle to Terminal 10% 13% +3% 30% Taxicab to Terminal 30% 30% 0% -2% Dropped Off at Terminal 30% 29% -1% -3% Transit to Airport 15% 15% 0% 0% Shared Van to Airport 5% 5% 0% -2% Scheduled Bus to Airport 5% 5% 0% -3% Total2 100% 100% Notes: 1 Percent difference calculations may differ due to rounding. 2 Totals may not add to 100% due to rounding. Source: Resource Systems Group, Inc., August 2009. Table 12. Addition of remote parking supply at a large-hub airport.

59 which would take several months to complete and additional commitment and investment by the airport operator. Since this is a specialized area, development of an airport-specific model requires specialized expertise. The decision to commission a new airport-specific model versus using the General Airport Park- ing Forecast Model will be based on the need to obtain results with a higher level of accuracy, with specificity to the airport, or with details not included in the General Airport Parking Fore- cast Model. The Final Report for ACRP Project 10-06 includes recommendations for data collection and enhancements to the model based on the research that the airport operator should consider when choosing between the General Airport Parking Forecast Model and development of an airport-specific model. One measure of the usefulness of the airport-specific park- ing forecast model is the assessment by the Port of Portland representative that the model results were similar to the results of the PDX Air Passenger Demand Model developed by the Port of Portland in 2009. Comparison of Airport-Specific and General Airport Models A comparison of the results of the airport-specific parking forecast model and the results of the General Airport Park- ing Forecast Model provides some insight into the value of developing an airport-specific parking forecast model. To evaluate the usefulness of the airport-specific model versus the general airport model, the results from each model with identical policy scenarios applied to the PDX environment were compared. Both models were calibrated to the specific characteristics of PDX. Two scenarios were developed to compare the results of the airport-specific and general airport models based on the specific characteristics of PDX. Scenario No. 1 tested a 50% increase in parking fees—from $30 to $45 for the “park and walk to termi- nal” mode and from $8 to $12 for the “park and ride shuttle to terminal” mode. Scenario No. 2 tested implementation of a $10 drop-off fee at curbside. Tables 13 and 14 present the results from both the airport-specific model and the general airport model. In reviewing these results, the differences should be con- sidered rather than the mode-share distributions. In the first scenario, the general airport model produces results that are similar to the PDX model. In the second sce- nario, the share of customers dropped off at the terminal (the customers who would be affected by this policy change) differs by 4 percentage points, which could indicate that customers in the 14-airport sample are generally less price sensitive than PDX customers or that they have fewer HOV options. More policies would have to be tested to compare differences in order to determine whether or not an airport operator should consider developing its own model or use the general airport model to test policy scenarios. Using Model Results to Estimate Impacts of Strategy Implementation Potential enhancements to the general airport model that would increase its usefulness in estimating the effects of con- strained airport parking include the addition of calculations of parking transaction parking revenue, vehicle trips generated by airline passengers, and related changes to vehicle emissions. The mode-share input and output from the general airport model can be used to estimate, at a high level, changes in vehi- cle trips and emissions by airline resident O&D passengers. The methodology described in Chapter 8 under “Measuring Effects of Parking Strategies” (in subsections on vehicle traffic Portland International Airport Mode Share Policy Scenario Access Mode Share Representative Base Case General Airport Model Airport-Specific Model Policy Scenario Difference1 Park and Walk to Terminal 10% 5% 4% 1% Park and Ride Shuttle to Terminal 15% 13% 12% 1% Taxicab to Terminal 10% 11% 11% 0% Dropped Off at Terminal 45% 50% 51% 1% Transit to Airport 10% 10% 11% 1% Shared Van to Airport 5% 5% 5% 0% Scheduled Bus to Airport 5% 6% 6% 1% Total 100% 100% 100% Note: 1 Policy scenario difference calculations may differ due to rounding. Source: Resource Systems Group, Inc., August 2009. Table 13. Comparison of general airport and airport-specific models with 50-percent increase in parking fees.

60 volume and emissions) can be used to determine the changes from implementing a strategy, as long as the other input data are available (such as vehicle occupancy by mode). However, some of the strategies may result in a change in the vehicle occupancy rate by mode, which is not predicted by the model. Parking facility exits (transactions) can be estimated for the two parking mode categories (i.e., “park and walk to terminal” and “parking and ride shuttle to terminal”), since exits are equivalent to private automobile trips in each category, with the qualification that a change in the vehicle occupancy rate by mode will influence the number of parking exits. It is not recommended that an airport operator use the model to esti- mate parking revenue because the model allows for only two parking categories (meaning two rates) and does not consider the average length of stay for parking customers (meaning that it does not consider changes to the average length of stay result- ing from different strategies). Some strategies will result in a change in the length of stay distribution by facility, which will affect revenues received. Informal Tools An airport operator may also use informal tools to estimate the effects of strategies being considered to address constrained airport parking. One approach is scenario analysis, which is the process of predicting, analyzing, and preparing for a range of effects associated with implementation of a variety of strate- gies to address constrained parking. The effects evaluated will be based on the airport operator’s goals and objectives related to the parking program. Based on estimates of changes in park- ing behavior at varying levels, effects evaluated may include revenues, vehicle trips generated by airline O&D passengers, and changes in vehicle emissions. The analysis may be based on experience, operational intuition, or benchmarks obtained from airports with similar operating environments and expe- rience. An example of a scenario analysis used to evaluate a strategy to address constrained parking would be estimating demand and revenue at different parking rates. Formal tools allow the user to look at a variety of outcomes related to passenger parking behavior and overall mode-share distribution, which is in many cases based on relationships established from underlying data. The user can compare strate- gies and understand the differences in outcomes at a certain level of reliability. The analysis results also may reveal outcomes that were unanticipated by the user. Informal tools are not as useful in comparing strategies because it is difficult to compare a range of potential results. In addition, changes in mode share will not be possible to predict because the relationships between mode preferences would not have been established. Benchmarks with the results from other airports also may provide some insights into the reasons for parking constraints and strategies that may relieve constraints and for making generalized comparisons with other airports. However, a wide range of variability typically can be found in these ratio-based benchmarks given the unique characteristics of each airport. Differing characteristics that may influence the ratio-based benchmarks include the percentage of airline passengers park- ing at an airport, the strength and availability of privately oper- ated off-airport parking, and, most importantly, whether the existing parking supply is adequate or is currently constrained, among other factors. Therefore, when airport operators con- sider which airports to benchmark, they should consider those with similar characteristics and benchmark against some with constrained parking and some without constrained parking. Benchmarks related to parking supply include the following: • Public parking spaces per O&D passenger—Because the majority of parking activity is generated by resident airline passengers, the ratio of parking spaces available per resident O&D passenger is a meaningful benchmark that attempts Portland International Airport Mode Share Policy Scenario Access Mode Share Representative Base Case General Airport Model Airport-Specific Model Policy Scenario Difference Park and Walk to Terminal 10% 12% 13% 1% Park and Ride Shuttle to Terminal 15% 19% 20% 1% Taxicab to Terminal 10% 12% 12% 0% Dropped Off at Terminal 45% 35% 31% 4% Transit to Airport 10% 11% 12% 1% Shared Van to Airport 5% 6% 6% 0% Scheduled Bus to Airport 5% 6% 7% 1% Total 100% 100% 100% Note: Totals may not add to 100% due to rounding. Source: Resource Systems Group, Inc., August 2009. Table 14. Comparison of general airport and airport-specific models with implementing a $10 drop-off fee.

61 to normalize the parking supply to potential customers. This information can be used to consider the supply needed in the future compared to forecast growth in number of air- line passengers. Other public parking ratios that may be con- sidered for different purposes include (1) airport-operated spaces per O&D passenger, (2) airport-operated spaces per resident O&D passenger, (3) total public parking spaces (airport operated plus privately operated) per O&D passen- ger, and (4) total public parking spaces (airport operated plus privately operated) per resident O&D passenger. • Composition of parking supply—Types of parking by O&D passenger or resident O&D passenger, or the per- centage of supply of a variety of parking products, such as long-term parking, short-term parking, or satellite park- ing, may provide insight. • Relationship between rates—The relationship between rates for different parking products may assist airport oper- ators in adopting rates that are different from rate changes made in the past. • Mode share—The nature of the airline passenger customer base, as well as the viable modes available to airline passen- gers based on service area, levels of service, and prices in relation to other modes will all influence the mode-share distribution at an airport. There may be value in compar- ing the airport’s mode share to mode shares at similar air- ports, but it is likely that the comparison will have less value than the other benchmarks listed. Table 15 presents a comparison of daily parking rates to other parking rates at the airports included in ACRP Project 10-06. Comparison to Daily Rates of Other Parking Products Airport Hub Classification1 Daily Parking Rate (Long-Term or Daily Parking Facility) Short-Term or Hourly Parking Valet Parking Economy Parking Boston Logan International (BOS) Large $24 – – 75% Chicago O’Hare International (ORD) Large $30 167% 150% 30%–53% McCarran International (LAS) Large $14 – 150% 57% Miami International (MIA) Large $15 200% 200% 53% San Diego International (SAN) Large $21 124% 143% 48%–76% Seattle-Tacoma International (SEA) Large $26 135% – – Tampa International (TPA) Large $15 133% 167% 60% Washington Dulles International (IAD) Large $17 212% 112% 59% Bob Hope (BUR) Medium $20 150% 100% 45%–55% Oakland International (OAK) Medium $22 145% 177% 68% Port Columbus International (CMH) Medium $17 159% 118% 35%–53% Portland International (PDX) Medium $14 171% 214% 57% San Antonio International (SAT) Medium $10 220% – 60% Huntsville International (HSV) Small $8 150% – 75% Tulsa International (TUL) Small $10 100% 180% 60% Notes: – means data are not applicable. 1 Hub size is defined by the FAA for commercial service airports based on the community’s share of total U.S. passenger boardings accommodated. Large-hub airports accommodate 1% or more of annual passenger boardings; medium-hub airports accommodate at least 0.25%, but less than 1% of passenger boardings; and small-hub airports accommodate at least 0.05%, but less than 0.25% of passenger boardings in the United States and its territorial possessions. Source: Ricondo & Associates, Inc. and DMR Consulting, based on airport case studies and representing conditions for different time periods (case studies collected from November 2008 through February 2009). (1–15) Table 15. Comparison of rates: relationship of daily parking rates to rates for other parking products.