Number of Independent Variable Studies Equation. Parking Spaces (PSP) 113 Ln(T) = Ln(PSP)

Size: px
Start display at page:

Download "Number of Independent Variable Studies Equation. Parking Spaces (PSP) 113 Ln(T) = Ln(PSP)"

Transcription

1 TripGeneration Characterislks of Shopping Centers BY JOAN C. PEYREBRUNE r is article presents the findings of a trip generation study performed for TE and the nternational Council of Shopping Centers (CSC) in 994 and 995. The study investigated several different aspects of trip generation of shopping centers, including: The relationship between trip generation and a combination of several independent H The definition and classification of shopping centers used by TE. The effects of the age of the data in the TE trip generation database. The relationship between pass-by trips and a combination of several independent Study Methodology and Statistical Analysis Procedures Two primary sources of data were used for the analysis: the TE trip generation database and data collected at 20 shopping centers by Raymond Keyes Associates for the CSC in 994. The TE database contained 556 studies in the Shopping Center Land Use Code 820, ranging in date from the 960s to 989. The database was supplemented with 20 additional studies conducted for the CSC. All of the analysis presented in this article is for the average weekday evening peak hour of adjacent street traffic, as this is typically the most critical time period in analyses of shopping center developments. The SAS/STAT statistical software was used to perform single and multivariable regression analysis based on the least-squares estimate methodology. The primary criterion for evaluating the results of the SAS analysis was the coefficient of determination (Rz). The R2 is the proportion of variance in one of the variables that can be explained by the variation in the other variable. The closer the R2 is to.0, the stronger the correlation. Additionally, the student t test and F ratios were used to compare mean trip rates and to test the significance of differences between groups for the analyses pertaining to the age of the data and shopping center classification. Trip Generation The primary purposes of the trip generation segment of this study were ) to develop a process and software program to perform multivariable analyses if it was determined to be appropriate, and 2) to further investigate trends in data that were already available. The sixth edition of Trip Generation, which is expected to be published in late 996, will provide revised trip generation rates with updated data. Based on preliminary analyses, it is not anticipated that the revised rates for shopping centers will be significantly different than those reported in the fifth edition. The trip generation analysis performed for this study is not intended to supersede the information provided in TE s Trip Generation, fifth edition. The single and multivariable regression analysis performed investigated the correlation between trips generated and the following independent variables: gross leasable area, peak hour adjacent street traffic, number of employees and parking spaces. The results of the analysis are shown in Table. As was expected, the analysis indicated that there was a high correlation between trips generated and gross leasable area (R2 = 0.80) during the evening peak hour. The number of trips generated had a low correlation (R2 = 0.2) with the evening peak hour adjacent street traffic. (t should be noted that it is believed that the adjacent street traffic may be a good predictor of trip generation for smaller retail stores, fast food restaurants, gas stations and other similar uses. The Trip Generation, fifth edition update, February 995, includes Table. Trip Generation Single Variable Analysis, p.m. Peak Hour of Adjacent Street Traffic of ndependent Variable Studies Equation R,,0~ Sq, Ft, Gross Leasable Area (GA) 365 Ln(T) = Ln(GLA) pm, Peak Hour Traffic on Adjacent Street 20 Ln(T) = 0,225 Ln(VOL) (VOL) Employees (EMP) 75 Ln(T) = 0,429 Ln(EMP) Parking Spaces (PSP) 3 Ln(T) = Ln(PSP) T = of p.m. peak hour trips generated 46 TE JOURNAL* JUNE 996

2 limited information on this variable; this study only investigated shopping centers [LUC 820].) The correlation with employees was relatively high (R2 = 0.57); however, this is not believed to be a good independent variable because the number of employees can vary greatly depending on the type of shopping opportunities and level of customer service available at the site. Another issue with using employees as an independent variable is typically the developer does not accurately know the number of employees at the time the traffic impact study is prepared. The correlation was also high for parking spaces; however, it was determined that this can primarily be explained because the number of parking spaces provided is typically based on the gross leasable area as dictated by local zoning ordinances. A further analysis indicated that the relationship between parking spaces and gross leasable area was linear with a correlation of 0.87, thus supporting this explanation. A multivariable analysis also was performed for the CSC data (20 studies) with all combinations of the independent The correlation of all combinations was relatively low. As discussed above, the primary reason for performing the multivariable analysis was to set up a process and develop a software program for this analysis. t was not felt that the variables analyzed would provide significant additional explanation in the variances however, there were no other data available to test the procedure, f additional data are collected, such as population, household income and other factors affecting the development market, the multivariable analysis may be a useful tool. The following recommendations are offered in regards to trip generation: The gross leasable area should be used as the independent variable, unless better information is known. Recognizing that gross leasable area does not explain all of the variability, Joan C. Peyrebrune, P. E., is a Transportation Engineer with Vanasse Hangen Brustlin nc. in Watertown, Mass. She was the Technical Projects Manager at the nstitute of Transportation Engineers in Washington, D. C., when this study was performed. - NEW DENOMNATOR TRAFFC COUNTERS AND TALLY BOARDS Featuring below-the-eounter positioning of keys, affording the lightest actuating pressure. R~gafized yst lightweight Lacquered maeonite boards-tailor made for your counting needs Work in ell kinda of weether Absolute accuraoy Write for broohure and price iat Ssrvingthosewhocountsince 94 THE DENOMNATOR CO., NC., WOODBURY, CT (203] 263=320 Time Switches (, 4 and 8 circuit) School Zone Flashers Solar &AC NEW!.arge Display Clocks & Timers (4, 8, 2 Bright LED Digits) Fire Station Warning Solar &AC Systems Detector Card Racks m RTC Emergency Vehicle Preemption Components High Water Detection and Warning Solar & AC Systems m Time Base Coordinators Solar Power Systems (4 and 8 circuit) Solar Panels & Components RTC Manufacturing, nc. -.--o P. O. Box 5089, Arlington, TX 7605 TE JOURNAL* JUNE

3 Table 2. Shopping Center Trip Generation Rates by Classification, p.m. Peak Hour of Adjacent Street Traffic Avg. l,o!! Sq. Ft. GLA Standard Reclassified TE 5th Percent of Studies of sample Deviation Rate Edition Rate Difference Neighborhood Center , % Community Center o% Regional Center , % Super Regional Center 79,048, % Trip rate per,colsq, ft GM. The rates were calculated using the regression equations and the avg, 0Xl sq, tt, GLA for each category consideration should be given to collecting additional data for various independent variables such as adjacent street traffic, population, household income, shopping opportunities and other socioeconomic Shopping Center Classification Currently there is one general land use category for shopping centers in the Trip Generation report (Land Use Code 820). This often causes confusion when users are trying to determine the appropriate land use category to apply to their study. n addition, many trip generation users believe there would be a significant change in the trip generation rates if the data were reclassified into different categories based on tenant type and size of development. The purpose of this section is to test this hypothesis by comparing the trip generation rates of Land Use Code 820 (Shopping Center) presented in Trip Generation, fifth edition, with revised rates calculated based on the reclassification of the data according to the definitions used by the Urban Land nstitute, as adopted by CSC. Trip Generation defines a shopping center as follows A shopping center is an integrated group of commercial establishments that is planned, developed, owned and managed as a unit. ts composition is related to its market area in terms of size, location and type of store. Shopping centers provide on-site parking facilities. Some of the centers included nonmerchandising uses such as office buildings, movie theaters, post offices, banks, health clubs and recreational facilities such as ice skating rinks or indoor miniature golf courses. Trip Generation goes onto state Separate equations have been developed for shopping centers less than and greater than 600,000 square feet gross floor area for the entire year excluding the Christmas shopping season, and for the Christmas shopping season. This methodology eliminates the need to categorize the center as a neighborhood, community, or regional shopping center, or by different land use codes other than 820. This is the first of several points of conflict in determining the trip generation of shopping center developments. The Urban Land nstitute emphasizes that a shopping center s type and function should be determined by its major tenant or tenants; they are never based solely on site area or square feet of the structure.z To test this hypothesis, all of the shopping center data were reclassified into one of the four categories below by reviewing the original data forms or studies submitted to TE. f these forms could not be located, an assumption was made based on the name of the development (for instance, plaza vs. mall ) or, as a last resort, on engineering judgment and intuition. t was decided that it was important to retain all of the data points so a true comparison could be made to Trip Generation, fifth edition, rates. The four categories of centers used were neighborhood, community, regional and super regional. Neighborhood Center. Provides for the sale of convenience goods (foods, drugs and sundries) and personal services (such as laundry and dry cleaning, barbering and shoe repairing) for the day-today living needs of the immediate neighborhood. t is built around a supermarket as the principal tenant. n theory, the neighborhood center has a typical gross leasable area of 50,000 square feet (sq ft); in practice it may range in size from 30,000 Sq ft to 00,003 Sq ft. Communiry Center. Provides a wider range of facilities for the sale of soft lines (wearing apparel for men, women and children) and hard lines (hardware and appliances), in addition to convenience goods and personal services. t is built around a junior department store, variety store or diseaunt department store as the major tenant, in addition to a supermarket. n theory, its typical size is 50,000 sq ft of gross leasable area, but in practice it may range in size from 00,000 sq ft to 450,000 Sqft. Regional Center. Provides for general merchandise, apparel, furniture and home furnishings in depth and variety, as well as a range of services and recreational facilities. t is built around one or two full-line department stores of generally not less than 00,000 sq ft. n theory, its typical size for definitive purposes is 450,000 sq ft of gross leasable area; in practice, it may range from 300,000 sq ft to 850,000 sq ft. Super Regional Center. Provides for extensive variety in general merchandise, apparel, furniture and home furnishings, as well as a variety of services and recreational facilities. t is built around three or more full-line department stores of generally not less than 00,000 sq ft each. n theory, the typical size of a super regional center is about 800,000 sq ft of gross leasable area; in practice, the size ranges from about 600,000 sq ft to more than.5 million sq ft. Table 2 shows the results of the regression analysis. The t -test and F -test cannot be used to compare the reclassified rates with the TE rates because the data are simply subsets of the same population. However, conclusions can be drawn based on the percentage difference in the rates. The entire traffic impact analysis procedure is based on several assumptions, such as the specific land use to be developed, the traffic distribution of the new traffic, the accuracy of the traffic counts of the existing traffic, the background growth rate, the percentage of development traftlc already traveling on the adjacent roadway, and the reliability of the computer model used for analysis. Keeping this in perspective, the small percent differences for the community center, regional center and super regional center should not be considered significant. The 2 percent difference for the neighborhood center could be considered significant. Consideration should be given to investigating these studies further to determine if they are properly classified and valid. The t -tests and F -tests were performed to determine if the reclassified 48 TEJOURNAL* JUNE 996

4 Table 3. Pm- and Post-980 Shopping Center Trip Generation Rates, Evening Peak Hour Standard of studies Rate Deviation Pre-980 Data and After Data Tripsper,CKClsquare feet gross leasable area rates were significantly different from each other. As expected, the rates were significantly different while comparing the neighborhood center, community center and regional center to each other. the regional center and super regional center rates were not significantly different from each other. Although the rates were different from each other, the current TE rates reflect this (as the size increases, the rate decreases). The following recommendations are offered in regards to shopping center classification The TE Land Use Code 820 definition should be expanded to note that all of the Urban Land nstitute classifications of shopping centers are included in the TE data to more accurately describe the data presented. Trip characteristics such as pass-by trips should be collected and separated for each subclassification to detect any trends between classifications. Consideration should be given to investigating the studies further to determine if they are properly classified and valid. Age of Data There has been much discussion regarding the age of the data contained in the TE trip generation database, which ranges from the 960s to 980s. t is apparent that travel patterns and behavior have changed during these three-plus decades. An analysis was performed to determine if the trip generation rates have changed significantly from before and after 980 by splitting the database into two subsets. As can be seen in Table 3, the rates are similar. Based on the t -test and the F-ratio, it was concluded that the mean trip rates were not significantly different. The following recommendations are offered: Retain all of the data to keep the database as large as possible. Undertake an extensive data collection effort to expand the database with more current studies. Pass-ByTrips Not all of the traffic generated by new developments is necessarily new to the adjacent roadway network; some of the trips may come from the traffic already passing the site (pass-by trips) and other trips may come from traffic on nearby roadways (diverted linked trips). Pass-by and diverted linked trips come directly from the traffic stream passing or near the development; they would be on the road whether or not the development was there. The percentage of pass-by and diverted linked trips may vary by the type and size of uses represented, the adjacent street traffic volumes, the time of day, the geographic location of the site relative to the urban center, and the nature of the roadway network serving the area, among other factors. The data analysis presented in this section will investigate the relationship of the percent of pass-by trips with the gross leasable CCD Cameras ntroduces the new Slopemaster Remote Series 4000 that will help you make the grade. Lenses Housings Domes Monitors E Pan/Tilts rhe new digital electronic inclinometer from Slope Meter is z.mique microprocessor based instrument designed to assist ir :he development and verification of slope and grade angles fol :he construction and engineering industries. Meeting bott ~eveiopment and verification objectives with one instrumen las resulted in an extremely versatile device capable of reduc ng both construction and inspection costs.,,,,* y Wide range of applications i Temperature compensated Easy to use and install Precise and accurate l ~~ Remote Sensing,:. ~. Compact design Serial interface., Dual axis Cable,Fiber Optic,PhoneLine, andwireless Transmission Equipment. Call or write for product information TEJOURNAL JUNE

5 Table 4. Percent Pass-By Trips Singie Variabie Anaiysis, p.m. Peak Hour of Adjacent Street Traffic of ndependent Variable Studies Equation R,,(KK Sq. Ft. Gross Leasable Area (GA) 20 Ln(T) = Ln(GLA) ,063 p,m, Peak Hour Tratfic on Adjacent Street 20 T = -0,00067 (VOL) 004 (VOL) + 28,833 Daily Traffic on Adjacent Street (ADD 20 T = (62.269/ADT ) 0.02 T= Percent pm-by trips area and the adjacent street p.m. peak hour traffic. Table 4 summarizes the results of the single-variable analysis. As can be seen, there is some correlation with the gross leasable area (R2 = 0.26) and virtually no correlation with the p.m. peak hour or daily traffic on adjacent street. As with trip generation, it is felt that the adjacent street traffic may be a good predictor in estimating pass-by trips for smaller retail stores, fast food restaurants, gas stations and other land uses with similar characteristics. The correlation was improved with the multivariable analysis (R2 = 0.54). A review of the partial R2 indicates that the adjacent street traffic explains an additional 4 percent. Figure presents a nomograph that can be used to predict the percent pass-by trips using both the gross leasable area and the p.m. peak hour adjacent street traffic. This figure is presented even though the correlation with adjacent traffic is low for this data set, because it is believed that similar nomography may be useful tools for other land use codes if data becomes available. Conclusions The trip characteristics of shopping gross leasable-area explains 50 percent of centers and other retail developments are the correlation, and the p.m. peak hour becoming more critical as communities focus on growth management and congestion relief. Therefore, it is important that complete and aeeurate impact studies be prepared to assess the impacts of these developments. To ensure that the traffic impacts of shopping centers are fully understood and investigated, the following final recommendations are offered, in addition to the specific recommendations made throughout this article: As always, additional data should be collected to further expand the TE database. Consideration should be given to collecting data for additional independent variables for both trip generation and pass-by trips. Consideration should be given to applying the methodology and procedures developed for this analysis to other land uses that may benefit from multivariable analysis. References. Trip Generation, fifth edition. Washington, D. C.: nstitute of Transportation Engineers, Dollars and Cents of Shopping Centers: 993. Washington, D. C.: Urban Land nstitute, ! g % 30?40 25% 20% 5% / ~ 2000 n # 000 n o o Sq. Ft. Gross Leasabie Area Fitted Curve Equation: %PASS-BY TRiPS = (X ) (X2) R2 = 054 Xl = P.M. Peak Hour Traffic on Adjacent Straet of Studies= 20 X2 = 000 Sq. Ft. Gross Leasebia Area caution should be used for sny development less than 200,000 sq. ft. GLA or grester than 800,000 sq. ft. GLA Figure. Shopping center pass-by trips during p.m. peak hour of adjacent street tratlic. 50 itejournal* JUNE 996