VEHICLE TRIP GENERATION FROM RETAIL, OFFICE AND RESIDENTIAL DEVELOPMENTS Dr D. O’Cinneide and R. Grealy University College Cork, Ireland 1. INTRODUCTION The traffic impacts of a new development depend mainly on the number of vehicles generated and on the capacity on the road and public transport networks. Consequently, each proposed development has unique characteristics. In applying for planning permission, traffic predictions typically use databanks such as TRICS (www.trics.org). These databanks are based on traffic counts at many existing developments and they are assumed to give reliable estimates of traffic volumes. However, little independent verification is available and substantial differences between the predicted and measured traffic volumes are not uncommon. In this paper, trip generation studies taken by students at University College Cork over a number of years are used to identify the principal factors which determine the volume of traffic from the three most common types of development in Ireland; retail food superstores, offices and residential developments. Comparisons are made with results derived from the TRICS databank. 2. TRAFFIC IMPACT ASSESSMENT Predicting the traffic impact of a proposed development involves: 1. Quantification of the local road and public transport networks; traffic flows, capacities, queue lengths, safety, etc. 2. Quantification of the number and distribution of the additional trips generated by the proposed development 3. Appraisal of the impact of these additional trips on the road and public transport networks Although trips by all modes of transport are included, vehicle trips normally contribute the most significant impacts. Based on the literature, the performance indicators shown on Table 1 are suggested for quantifying the traffic impacts of a proposed development. The environmental impacts resulting from additional traffic are not included. Methods of quantifying each performance indicator are also indicated in Table 1. These methods use UK traffic models (www.trlsoftware.co.uk; www.sias.com) since studies carried out at University College Cork have indicated that UK traffic models are applicable in Ireland (Grealy 2006, O’Cinneide and O’Mahony 2005). © Association for European Transport and contributors 2008 1 Table 1: Performance indicators for measuring the traffic impact of developments Traffic impact Performance indicator Method of measurement parameter Area and individual junction Congestion Queue lengths: Predicted simulation models such as changes in queue lengths at Paramics, OSCADY, significant junctions ARCADY Delay: Predicted changes in delay at significant junctions Junction models such as Safety Number of serious and minor OSCADY or ARCADY and/or network accident prediction injury accidents: Predicted models such as SafeNET2 changes in the number of serious and minor accidents on adjacent road links and junctions. Vulnerable road users Predicted change in delay when crossing at significant junctions Area and individual junction models such as Paramics and OSCADY The performance indicators shown on Table 1 depend on the type and size of a proposed development, on the traffic generated and on the surrounding road network before and after the development. Since network conditions are site specific, this paper only considers the trip generation from developments. 3. THE SELECTION OF LAND USES FOR DETAILED INVESTIGATION The main criteria for the selection of the land uses for detailed study were: 1. The most common types of developments in Ireland. 2. Ease of obtaining the required site information, traffic counts, etc. In recent years most new developments in Ireland can be categorised under land use categories: Retail, Office Block and Residential or as a mix of these land use categories. Each category generates different types of trips (work trips, commercial trips, home trips, etc.). Also, most of the trips from these land uses are distributed to the other two land uses. The locations are also influenced by the locations of the other two categories. © Association for European Transport and contributors 2008 2 4. TRIP GENERATION CASE STUDIES 4.1 Traffic Counts at Selected Land Uses in the Cork Region The principal traffic impacts from a new development occur during peak traffic periods on the adjacent road network. Additional vehicles have a particularly severe effect if the network has exceeded capacity. Consequently, the traffic counts described in this paper are the number of vehicles by type entering and leaving existing developments during the morning (8.00-9.00) and / or evening (17.00-18.00) peak traffic periods. These counts were taken at various times from 2004 to 2007. For further details see Grealy 2006. 4.2 Retail Development Counts These were restricted to food stores only. Food stores are divided into two subcategories; food superstores and discount food stores because substantially higher trip rates were expected from the later. Food superstores vary from standalone developments, to those which include non-food elements, and to stores which are part of larger shopping centres. A number of different sizes and types of food store in the Cork area were selected for trip rate analysis. Table 2 shows the details and the measured light vehicle trip rates (per 100m2 gross floor area) for each development studied. Apart from Tesco Wilton, which remained open for 24 hours, these stores opened at 9.00 hours and closed after 19.00 hours when the counts were taken. Counts were limited to suburban food stores because of the difficulty in quantifying vehicle trip rates for city centre food stores. • Food superstores with non-food sections The total evening peak hour (17.00-18.00) trip rates for Dunnes, Bishopstown Court and for Dunnes Stores Ballyvolane were fairly similar. These are similar food superstores with similar characteristics regarding the surrounding road networks and the population within a 17- minute travel time. • Food superstores which are part of larger shopping centres The total evening peak hour (17.00-18.00) trip rate for Wilton Shopping Centre was marginally higher than at the slightly smaller Douglas Court Shopping Centre. These shopping centres have similar populations within a 17 minute travel time and the Cork South Ring Road (N25) is located adjacent to both shopping centres. Both developments were similar with each having banks, restaurants and commercial units. • Stand alone food superstores / discount food stores The measured evening peak trip rates were substantially higher than for food superstores with non-food sections or for superstores in shopping centres. © Association for European Transport and contributors 2008 3 The total evening peak hour trip rate for Lidl, Togher in 2004 was 21.96 veh/100m2 while counts at the same development in 2002 showed a total evening peak hour trip rate of 19.44 veh/100m2. Table 2 Food superstores in the Cork area selected for trip rate analysis Name Location and type Size Time Trip rates (light m2 vehicles per 100m2 GFA* GFA* In Out Total 1.22 0.25 0.97 7189 8-9 Suburban Dunnes, Bishopstown foodstore with non8.71 3.91 17-18 4.80 food section Court, Cork 6520 Suburban Dunnes, Ballyvolane, foodstore with non9.50 4.65 17-18 4.85 food section Cork 3.03 1.33 1.70 Suburban 16462 8-9 Dunnes, shopping centre Douglas 7.62 4.49 17-18 3.13 Court Centre 3.90 1.95 1.95 Suburban 17656 8-9 Tesco shopping centre Wilton 7.86 4.63 17-18 3.23 Centre 1.94 0.51 1.43 Satellite town 1960 8-9 SuperValu foodstore Ballincollig, 20.66 17-18 11.02 9.64 Co. Cork 1.84 0.34 1.50 Lidl, Togher, Suburban discount 1471 8-9 Cork foodstore 17-18 11.08 10.88 21.96 Co. Cork town 1610 Lidl, discount foodstore Midleton, 16.89 8.57 17-18 8.32 Co. Cork * GFA = Gross Floor Area; the sum of the areas of all the floors of a building • Relationship between size of food stores and trip rates From Table 2 it can be seen that a decrease in the size of a development normally results in an increase in the total evening peak hour trip rate. SuperValu, Ballincollig (a satellite town food store) was only 27% the size of Dunnes, Bishopstown Court. However, the total evening peak hour vehicular movement at SuperValu Ballincollig was 65% of that at Bishopstown Court. The substantially higher trip rate was partly due to the lack of competition in the surrounding area; the only competition was a discount food store at the opposite end of the town. © Association for European Transport and contributors 2008 4 Table 2 shows that that the similar sized Wilton and Douglas Court Shopping Centres have almost the same total evening peak hour trip rates. Wilton and Douglas Court Shopping Centres have: o Similar populations within a 17-minute travel time. o A similar development located within a mile. o The Cork South Ring Road (N25) is adjacent The gross floor area of Lidl, Togher is 8.6% smaller than that of Lidl, Midleton but the total evening peak hour trip rate for Lidl Togher was 30% higher than that for Lidl, Midleton. The probable reason for Lidl Togher having a higher evening peak hour trip rates is the adjacent population. The population within a mile of Lidl, Midleton is substantially less than that of Lidl, Togher. From the studies of the food superstores, three general conclusions are drawn: 1. The lack of similar developments in an area result in high trip rates 2. The smaller the foodstore, the higher the trip rate. 3. Similar developments with similar populations, road networks and competition have similar evening peak hour trip rates. 4.3 Office Block Development Counts The office blocks selected for analysis are shown on Table 3. The gross floor area (GFA) in this table includes occupied office blocks only. Table 3. Offices in the Cork area selected for trip rate analysis Name Location Type Size Time Trip rate (light m2 vehicles / 100m2) GFA In Out Total 1.49 0.09 1.58 Suburban Single 4393 8-9 Jacobs company Engineering, 17-18 0.15 1.53 1.68 Mahon, Cork Suburban 4 office 11000 8-9 1.45 0.12 1.57 University blocks Technology Centre, Curaheen Rd., Cork 3.33 0.14 3.47 RPS-MCOS, Satellite Single 1408 8-9 Ballincollig, town company 17-18 0.07 1.56 1.63 Occupied 37161 8.00- 1.63 0.10 1.73 Cork Airport Cork 9.00 Airport office Business blocks Park © Association for European Transport and contributors 2008 5 Two trip rate parameters are used to express office block trip rates; the size of the office block and the number of employees. When using size as a trip rate parameter, the entire development must be in operation to ensure a useful measured trip rate. The number of employees would appear to be a suitable trip rate parameter, particularly if most employees remain in the office all day and work similar hours. A detailed investigation into the two engineering consultancy companies studied (Jacobs Engineering and RPS-MCOS) showed that both had similar morning vehicle peak hour trip rates per employee; Jacobs Engineering (166 employees) had a trip rate of 0.56 light vehicles per employee compared with 0.43 at RPS-MCOS (88 employees). However, the evening peak hour trip rate per employee at Jacobs Engineering (0.46) was substantially higher than at RPS-MCOS (0.26). This difference resulted mainly from the proportions of the employees traveling to external sites during the day and from the proportions working beyond 18.00 hours. This suggests that the number of employees may not be a suitable trip rate parameter. Also, it is difficult to predict the number of employees in a proposed office block development. Consequently, the trip rates shown in Table 3 are in terms of trips per 100 m2 of gross floor area. From Table 3 it can be seen that the morning peak hour trip rates per 100 m2 gross floor area were fairly similar for 3 of the 4 office blocks studied. However, the trip rate at RPS-MCOS was about twice that of the other blocks. Only part of this difference was due to employees arriving before the morning peak hour. It is difficult to draw reliable conclusions from these office block studies. The differences in the measured vehicle trip rates could not be explained by differences in the modal split (almost all the employees used a car), or by differences in arrivals/departures before or after the peak hours. The type of business being carried out appeared to significantly affect the trip rates. 4.4 Residential Development Counts The most common types of residential developments in Ireland are single family isolated dwellings in rural areas, suburban or ex-urban housing estates with 3/4/5 bed dwellings or apartment blocks in cities and towns. The residential developments studied and the measured morning peak hour (8.00-9.00) car trip rates are shown on Table 4. The number of heavy vehicles was minimal. The measured car trip rates for the morning peak hour varied from 0.27 to 1.65 trips/household; 75% - 80% of the vehicles were departing. The principal factors which influenced the morning peak hour trip rates were the type and number of occupants (wealth, employment, age), the car ownership level and the location. The social status and employment of the occupants were strongly related to the car ownership levels. The location of a residential development influenced both the type of occupant and the car ownership level. © Association for European Transport and contributors 2008 6 Table 4. Residential development morning peak hour (8.00-9.00) trip rates. Name No. of Type Car trips per households household In Out Total Broadale, Maryborough 356 4-5 bed semi + 0.18 0.73 0.91 Hill, Cork detached Knockrea Park, 58 3 + 4 bed semi 0.62 0.29 0.91 Ballinlough, Cork + detached Heatherfield, Waterfall, 43 5 bed detached 0.00 1.08 1.08 near Cork 0.25 0.86 1.16 Ferndale, Carrigaline, 74 3 / 4 bed semi 0.13 1.14 1.27 Co. Cork detached Park Gate, Frankfield, 295 3 / 4 bed semi 0.89 Cork detached Mount Oval, 26 Detached 0.26 0.86 1.12 Rochestown, Cork Shamrock Lawn, 489 3 / 4 bed semi 0.93 Douglas, Cork detached Deerpark, Cork 98 Town houses 0.09 0.85 0.94 Haldene, Bishopstown, 64 3 / 4 bed semi 0.18 0.74 0.92 Cork detached Willowbank, Church Rd. 103 Semi detached 0.31 0.80 1.11 Blackrock, Cork Cloghroe, Co. Cork 29 Semi detached 1.17 Outskirts of Bantry, Co. 25 Mix detached + 0.20 0.52 0.72 Cork semi detached Norwood, Cobh, Co. 180 3 / 4 bed semi n/a n/a 0.76 Cork detached Carraigroe, 67 Apartments 0.10 0.27 0.37 Mitchelstown, Co. Cork (immigrants) Oakwood, Macroom, 110 3 / 4 bed terrace 0.05 0.22 0.27 Co. Cork (immigrants) Woodfield, Killeagh, Co. 55 10 semi + 45 0.07 0.43 0.45 Cork detached Ballinacurra, Midleton, 56 Detached 0.00 0.45 0.45 Co. Cork Crookstown, Co. Cork 65 Semi detached 0.42 Woodside, Ballynora, 13 Detached 0.23 0.77 1.00 Cork Mardyke Walk, Cork 59 Mix of house 0.53 types Culrua, Portlaw, Co. 81 Semidetached 0.59 0.22 0.81 Waterford 26 Semi +detached 0.40 1.25 1.65 Carystown Woods, 83 Mix of house 0.15 0.64 0.78 Whitegate, Co. Cork types © Association for European Transport and contributors 2008 7 Table 5 shows the median and range of the total morning peak hour trip rates for the observed developments divided by location into Inner City / Town, Cork City Suburbs, Rural Areas on the edge of Cork City, and County Cork towns. Inner City: Areas within a feasible walking distance (<1km) of the city/town centre Suburbs: Areas within a feasible walking distance of a Cork City Bus Service but outside a feasible walking distance of the city centre Rural Areas: Areas with little public transport service and outside a feasible walking distance of the city centre Table 5 Total morning peak hour (8.00-9.00) car trip rates by location Location Median car trip Range rate / household Inner city / 0.53 0.27 – 0.94 town Suburbs of 0.94 0.89 – 1.11 Cork City Rural areas 1.11 1.00 – 1.27 County towns 0.69 0.42 – 1.65 Morning peak hour trip rates for inner city / town developments were generally lower than for suburban or rural developments. The ratio of cars/household was well below 1.00 for all but one of these inner city developments (Deerpark). Many occupants were retired people or tenants with restricted budgets such as recent immigrants. 5. THE ACCURACY OF EXISTING METHODS OF TRIP RATE CALCULATION The most frequently used database for predicting trip rates in Ireland and the UK is TRICS (www.trics.org) which is a large database based on traffic counts from a large number of sites throughout Great Britain and Ireland and is updated regularly. However, most data is from the south east of England. The TRICS land use categories are divided into 16 main categories which are further divided into sub-categories. Trip rates for each sub-category type are given as the number of vehicle or person arrivals and departures per 100m2 gross floor area, number of employees, households, etc., and can be used to predict the traffic/people flows in and out of proposed developments by half hour of the day and day of the week. TRICS trip rates are compared with the measured trip rates in the following pages. © Association for European Transport and contributors 2008 8 5.1 Comparison between Observed and Predicted Trip Rates for Retail Developments Table 6 shows the calculated TRICS trip rates and the measured trip rates for the evening peak hour (17.00 -18.00) for the food stores studied. Table 6. TRICS trip rates and measured rates for food stores (17.00 -18.00) Name Location and Trip rates (light vehicles per 100m2 type GFA In Out Total Suburban Dunnes, Observed 4.80 3.91 8.71 Bishopstown foodstore with TRICS 6.47 6.48 12.95 non-food Court, Cork section Suburban Dunnes, Observed 4.85 4.65 9.50 Ballyvolane, foodstore with TRICS 6.47 6.48 12.95 non-food Cork section Dunnes, Douglas Court Centre Tesco Wilton Centre Suburban shopping centre Suburban shoping centre Observed TRICS 3.13 4.49 7.62 6.32 Observed TRICS 3.23 4.63 7.86 6.32 SuperValu Ballincollig, Co. Cork Satellite town foodstore Observed TRICS 11.02 3.32 9.64 3.64 20.66 6.96 Lidl, Togher, Cork Suburban discount foodstore Co.Cork town discount foodstore Observed TRICS 11.08 3.32 10.88 3.64 21.96 6.96 Observed TRICS 8.32 3.32 8.57 3.64 16.89 6.96 Lidl, Midleton, Co. Cork For food superstores with non- food sections, the gross floor area used for TRICS was similar to that of the foodstores studied. However, the predicted trip rates were considerably higher than the measured trip rates. For food superstores which are part of larger shopping centres (Douglas Court Shopping Centre and Wilton Shopping Centre), TRICS did not include any information in the food superstore sub-category for developments of similar size. © Association for European Transport and contributors 2008 9 However, developments categorised by TRICS under retail parks - including food have similar gross floor areas. Assuming this category, TRICS underestimated the peak hour trip rates for both shopping centres. For stand-alone food superstores, TRICS did not include this sub-category for developments of similar size to SuperValu, Ballincollig. However, developments categorised by TRICS under discount food stores have similar sizes to SuperValu, Ballincollig. Assuming the discount food stores category, TRICS predicts a total evening peak hour trip rate of 6.96 vehicle trips/100m2 for an average gross floor area similar to SuperValu, Ballincollig. This underestimates the measured evening peak hour trip rate by 13.70 trips/100m2, which is substantial. For equivalent size discount food stores, the evening peak hour trip rate predicted by TRICS substantially underestimated the measured trip rates. In summary: • For food superstores with non-food sections using TRICS overestimated the evening peak hour trip rates. • For food superstores which are part of larger shopping centres, the predicted TRICS rates were less than the measured trip rates. • For stand alone food superstores and discount food stores, the measured trip rates were considerably greater than the trip rates predicted using TRICS. The differences between the measured trip rates and those predicted using TRICS could be partly because TRICS did not have exact equivalent categories to the retail food superstores studied in the Cork region. 5.2 Comparison between Measured and Predicted Trip Rates for Office Blocks Table 7 shows the predicted TRICS trip rates and the measured trip rates for office blocks. From this table it can be seen that TRICS overpredicted the morning peak hour (8.00-9.00) trip rates for three of the Cork office blocks and underpredicted the other. However, these differences are reduced when the trip rates from 7.30 – 8.00 are taken into account. The evening peak hour trip rates using TRICS were similar to the trip rates at the two office blocks where evening peak hour counts were taken. © Association for European Transport and contributors 2008 10 Table 7. Predicted and measured peak hour trip rates for office blocks Name Time Trip rates (light vehicles per 100m2 GFA) IN OUT TOTAL Jacobs 7.30-8.00 Observed 1.08 0.08 1.16 Engineering, TRICS 0.35 0.06 0.41 Mahon, Cork 8.00-9.00 Observed 1.49 0.09 1.58 TRICS 1.94 0.21 2.15 17.00-18.00 Observed 0.15 1.53 1.68 TRICS 0.24 1.53 1.77 8.00-9.00 University Observed 1.45 0.12 1.57 Technology TRICS 1.46 0.25 1.71 Centre RPS-MCOS, 7.30-8.00 Observed 0.78 Ballincollig, TRICS 0.35 0.06 0.41 Cork 8.00-9.00 Observed 3.33 0.14 3.47 TRICS 1.94 0.21 2.15 17.00-18.00 Observed 0.07 1.56 1.63 TRICS 0.24 1.53 1.77 Cork Airport 8.00-9.00 Observed 1.63 0.10 1.73 Business Park TRICS 1.82 0.18 2.00 5.3 Comparison between Measured and Predicted Trip Rates for Residential Developments Table 4 showed the measured total morning peak hour trip rates for residential developments in the Cork region. The total morning peak hour trip rates varied from 0.27 cars trips/household to 1.65 car trips /household. Median trip rates for Inner City, Suburban, Rural and County Town developments were shown in Table 5. Table 8 compares these median trip rates with the equivalent predicted values using TRICS. Table 8 Predicted and measured total morning peak hour (8.00-9.00) car trip rates by location Location Median total car Range TRICS total car trip trip rate/ household rate / household Inner city / 0.53 0.27 – 0.94 Privately owned: 0.11 town Rented: 0.08 Suburbs of 0.94 0.89 – 1.11 0.62 Cork City Rural areas 1.11 1.00 – 1.27 0.70 County towns 0.69 0.42 – 1.65 0.70 The TRICS predicted trip rates are substantially lower than the measured trip rates for than the measured trip rates for inner city, suburban and rural residential developments. © Association for European Transport and contributors 2008 11 6. DISCUSSION The traffic impact of a proposed development depends on the type and size of the development, on its location and on the receiving road network. As previously indicated, the impact of a specified amount of vehicular traffic on the road network can be quantified within an acceptable level of accuracy using established United Kingdom junction and network traffic models. 6.1 Retail Food Superstore Trip Generation Case Studies Peak hour trip rates were measured for two food superstores with non-food sections, two food superstores which are part of larger shopping centres, one stand-alone food store and two discount food stores. These developments were all located in the Cork City region. The measured trip rates indicated that the peak hour trip rates varied by the type of food store but were consistent for each type of food superstore. The evening peak hour trip rates for the stand-alone food superstore and the discount food stores were substantially higher than for the other types of food superstore studied. The measured peak hour trip rates for retail food superstores did not compare well with the rates predicted using the TRICS database. It is suggested that this may be because TRICS does not include a sufficiently detailed breakdown of the different types of food store (the retail food superstore categories in TRICS were not directly comparable with the food superstores studied). Also, TRICS does not include whether there are similar developments in the vicinity. Consequently, it is suggested that the traffic from a proposed retail food superstore can only be accurately estimated by comparison with existing developments in similar locations and distances from competitive developments. 6.2 Office Block Trip Generation Case Studies The morning peak hour trip rates were expressed in terms of trips per 100m2 gross floor area rather than per number of employees which was not considered a suitable parameter. Based on the limited studies of office blocks in the Cork City region, it is suggested that the size of an office block has little influence on the morning and evening peak hour trip rates per 100m2. However, the operating hours of offices appeared to have a significant influence. At present TRICS does not include information on office operating hours. The trip rates predicted by using TRICS were fairly similar to the trip rates measured during the case studies. Thus it is considered that the TRICS database reasonably represents the trips generated by offices. © Association for European Transport and contributors 2008 12 6.3 Residential Trip Generation Case Studies The residential development case studies indicate that the location of a development has a significant impact on the morning peak hour trip rate. The morning peak hour car trip rates predicted by using the TRICS database were significantly lower than the measured rates. This may partly result from lower public transport use in Ireland than in the UK. To predict accurate residential morning peak hour trip rates, the location relative to the city centre should first be identified (Inner City, Suburban, Rural, County Town). Developments in similar locations should then be used to estimate the morning peak hour trip rates. 7. CONCLUSION This paper investigates the traffic impact assessment of developments in Ireland. Trip rates from existing developments were measured mainly in Cork City and County Cork. Conclusions include that the traffic impacts of a proposed development depend primarily on the number of trips generated and that there was a wide range of trip rates for the developments studied. Also, that existing methods for estimating trip generation were not accurate for certain types of development. In Ireland, the TRICS database is typically used for trip rate calculation when applying for planning permission. Reduced trip rates are then assumed where new public transport services or mobility management plans are proposed. TRICS was found to give reasonably accurate trip rates for office developments but was not considered appropriate for retail food superstores or for residential developments. This suggests that the impacts of the latter developments require extra attention and should not rely solely on the TRICS database. Traffic impact assessments depend directly on the assumptions made about developments. These assumptions are often biased since those seeking planning permission tend to make favourable assumptions. Post development traffic impact assessments are required to verify the accuracy of the process. Bibliography Grealy, R. (2006). Traffic impacts of developments. MEngSc thesis. University College Cork. O’Cinneide, D and B. O’Mahony (2005). The evaluation of traffic microsimulation modelling. Urban Transport XI. Urban Transport and the Environment in the 21st Century, WIT Press Southampton. TRICS www.trics.org. © Association for European Transport and contributors 2008 13