UNSW Sydney SEMESTER 2 2017 EXAMINATIONS CVEN9405: Urban Transport Planning Practice 1. TIME ALLOWED – 2 hours 2. READING TIME – 10 minutes 3. THIS EXAMINATION PAPER HAS 12 PAGES 4. TOTAL NUMBER OF QUESTIONS – 6 questions in two sections 5. ANSWER THE QUESTIONS OF EACH SECTION IN A SEPARATE ANSWER BOOKLET 6. TOTAL MARKS AVAILABLE – 100 7. MARKS AVAILABLE FOR EACH QUESTION ARE SHOWN IN THE EXAMINATION PAPER 7. ALL ANSWERS MUST BE WRITTEN IN INK. EXCEPT WHERE THEY ARE EXPRESSLY REQUIRED, PENCILS MAY BE USED ONLY FOR DRAWING, SKETCHING OR GRAPHICAL WORK 8. THIS PAPER SHOULD BE RETURNED Section 1 Question 1: [10 Points] Determine which one of the following statements is true and which one is false. a) When converting number of trips from person-trip to vehicle-trips, the vehicle occupancy rate should be rounded to the nearest integer as decimals are not meaningful. (2 points) b) When determining Traffic analysis zone (TAZ) boundaries in urban transport planning studies, TAZs should have identical size. (2 points) c) Accessibility is determined based on transport infrastructure. (2 points) Solution Statement π) π) π) True/False False False True Question 3: [20 Points] The following data is available from a travel survey in a city with 6 zones. Zone Trip generation (Y) Population (πΏπ ) 1 2 3 4 5 6 Mean 209 199 210 190 202 156 194.3 108000 94000 115000 101000 94000 78000 98333.3 Number of employees in industrial jobs who have access to car (πΏπ ) 20758 20038 20898 17192 20580 15239 19117.5 Assume the following three linear regression models are candidate trip generation models for this region. Model ID Model specification (i) (ii) (iii) π = π½0 + π½1 π1 π = πΌ0 + πΌ2 π2 π = π0 + π1 π1 + π2 π2 The results of regression analysis for each model are provided in the next page. Based on the results, answer the following questions. a) Compare the goodness-of-fits in models. Explain which model better fits the data and specify the measurement that you use for the comparison. (4 points) b) Considering a confidence level of 95 per cent, discuss the statistical significance of coefficients in model (iii). Specify the values that you use for this purpose. (4 points) c) Which model would you select as the best model for trip generation prediction purposes? Explain your criteria and justify your answer. (6 points) d) Calculate trip generation for a zone with the mean values of π1 and π2 using each of the three models? Explain your answer and show all the work. (6 points) Model (i) Regression Statistics Multiple R R Square Adjusted R Square Standard Error Observations 0.861 0.742 0.677 11.445 6 Intercept Population (X1) CoefficientsStandard Error t Stat 61.85209 39.35941 1.571 0.001347 0.000397 3.390 P-value 0.191 0.028 Model (ii) Regression Statistics Multiple R R Square Adjusted R Square Standard Error Observations 0.949 0.900 0.875 7.131 6 CoefficientsStandard Error t Stat 38.83076 26.11147 1.487 Intercept Number of employees in industrial jobs who have access to car (X2) 0.008134 0.001357 5.993 P-value 0.211 0.004 Model (iii) Regression Statistics Multiple R R Square Adjusted R Square Standard Error Observations Intercept Population (X1) Numebr of employees in industiral jobs who have access to car (X2) 0.983 0.966 0.943 4.812 6 CoefficientsStandard Error t Stat 25.58579 18.46185 1.386 0.000579 0.000241 2.405 P-value 0.260 0.095 0.005848 0.00132 0.021 4.430 Solution a) Since the models have different number of coefficients, a suitable measurement to compare their goodness-of-fit is adjusted R square. According to the results, model (iii) has the highest adjusted R square (0.943). After that comes model (ii) with an adjuster R square of 0.875 and finally model (i) with 0.677. b) This model has two explanatory variables of π1 and π2 . According to the p-value, the coefficient of π2 is not statistically significant with 95 percent confidence (because 0.095 is greater than 0.05), while the coefficient of π1 is statistically significant. c) Model selection criteria should include: goodness-of-fit, statistical significance of coefficients, and number of explanatory variables. • Model (iii) has a non-significant coefficient (and probably multicollinearity) so it should not be selected • The decision between model (i) and (ii) boils down to the tradeoff between achieving higher goodness-of-fit and the difficulty of predicting future values for explanatory variables. • Students can argue either way and as long as they have acknowledged the trade-off they get the full mark in this section d) Regression line goes through the mean values. So the answer will be 194.3 for all the models. Students get full mark if they explain this fact or if they do the correct calculation for all the models. Section 2 Question 4: Trip Distribution [20 marks] a) (10 marks) i) ii) In one sentence describe the concept of trip distribution in the context of the 4-step transport model? (3 marks) Trip distribution matrices can be presented in two formats, what are these formats? What is the difference between the two methods? (4 marks) Answer True or False for the following (3 marks) iii) iv) v) Growth models (e.g., Fratar Model) used for trip distribution are able to make provision for zones which are at present undeveloped. (T/F) Gravity models used for trip distribution are able to make provision for zones which are at present undeveloped. (T/F) Trip distribution can be completed after mode choice in the 4-step transport model. (T/F) Solution i) ii) iii) iv) v) Trip distribution is the second step of the 4 step transport model and involves determining the origins and destinations of the generated trips from the first step. (3 marks) Origin-Destination Matrix, Production-Attraction Matrix (2 marks). OD-Matrices account for trip direction while production-attraction matrices do not account for direction (are based on land use alone) (2 marks) F (1 mark) T (1 mark) T (1 mark) b) (10 marks) Approximately 80,000 people work in a city centre. In general, these workers live in four neighbouring suburbs within 20 miles of the city centre. The number of households and travel time from each travel zone to the city centre is shown below within part i) of the question. The following function defines the gravity model used to distribute trips within the region for a particular time of day, neglecting intra-zonal trips. π΄π 2 π‘ππ πππ = ππ ( ∑ππ=1 π΄π 2 π‘ππ ) πππ = number of trips generated in zone π that terminates in attraction zone π ππ = number of trips originating in zone π π΄π = number of trips attracted to zone π π = number of zones π‘ππ = travel time between zone π and zone π (assume that π‘ππ = π‘ππ ) i) Complete the following table and determine the total values for each column in the final row. State any assumptions. (5 marks) Suburb Total Attractions to each suburb (π΄π ) Travel time (π‘ππ ) 1 2 3 4 25,000 15,000 30,000 37,500 30 15 9 18 π΄π 2 π‘ππ π΄π 2 π‘ππ π΄ ∑ππ=1 2π π‘ππ πππ Total ii) iii) Solution i) Describe in one sentence what the above table calculates from the perspective of trip distribution. (2 marks) Does the total of the “Total Attractions to each suburb” column equal the total value of the πππ column? If not, which column had a greater total and why? (3 marks) Assume the number of trips originating in the city during an afternoon peak period: Ti = 80,000 (1 mark, 1 mark for each column correctly calculated, ½ marks for partially correct columns) Suburb A B C D Total Total Attractions to each suburb (π΄π ) Travel time (π‘ππ ) 25,000 15,000 30,000 37,500 107,500 30 15 9 18 π΄π 2 π‘ππ π΄π 2 π‘ππ ∑ππ=1 ∑ π΄π 2 π‘ π‘ππ = 27.78 66.67 370.37 115.74 580.56 0.048 0.115 0.638 0.199 1 ∑=1 ii) iii) πππ π΄π 2 π‘ππ 3,828 9,187 51,036 15,949 80,000 ∑ = 80,000 The above table calculates the number of trips originating from the city zone that are destined to suburb zones A, B, C and D, reflecting afternoon commuter trips. (1 mark for stating where trips originate, 1 mark for stating the destination) No the column values are not equivalent. The “Total Attractions to each suburb” has a value of 107,500 which is greater than the πππ column which totals to 80,000. This is the case as the πππ column represents trips originating from only the city zone; there would be trips between each suburb which would form the entire trip distribution of the region. (1 mark for stating ‘No’, 1 mark for indicating which column had a greater value, 1 mark for reasoning) Question 5: Mode Choice [20 marks] a) (10 marks) i) Using the graph below, determine the number of public transport trips per day in a zone which has 5000 people living on 50 hectares. The private vehicle (autos) ownership statistics reveal that 40% of the households do not own a private vehicle and the remaining households own 1 vehicle per household. (5 marks) ii) There are numerous factors which influence mode choice, please list 5 of the factors that you believe affect mode choice. (5 marks) Solution i) ii) • Number of persons per hectare = 5000/50 =100 (1 mark) According to the graph: 0 autos/HH – 510 trips/day/1000 population (1 mark) 1 auto/HH – 250 trips/day/1000 population (1 mark) Total public transport trips = 0.4*510*5 +0.6*250*5 = 1020+750 = 1770 trips per day (2 marks) Note that 510 and 250 are simply read from the graph, so it’s fine that someone reads, for example, 500 and 200. (1 mark for each correctly identified factor) Characteristics of the trip maker: o Car availability o Possession of a driving licence o Age/Household structure o Income o Residential density/Accessibility Characteristics of the journey o Trip purpose o Time of the day o Travelling alone or in groups Characteristics of the transport facility o Travel time o Fares and monetary costs o Parking costs and availability o Reliability of the mode o Comfort o Safety • • b) (10 marks) Tourists to a particular region take advantage of three exotic modes of transport at a resort; helicopter, hovercraft and balloon. The utility function is given by ππ = ππ − (0.009 × πΆππ π‘) − (0.057 × πΌπππ) − (0.061 × ππππ) where πΆππ π‘ is the price in dollars of the ticket, πΌπππ is the in vehicle travel time and ππππ is the out of vehicle travel time (time spent waiting). The variables for the three modes are as follows, where π provides a subscript identifier for each mode. π mode ππ Cost IVTT OVTT 1 2 3 Helicopter Hovercraft Balloon 0.0 0.4 1.1 85.40 88.16 70.59 5.35 15.7 38.0 33 34 42 Answer the following questions, considering a multinomial logit model used to describe mode choice which has the following general functional form: π(πΎ) = π ππ ∑ππ₯=1 π ππ₯ Where: π(πΎ) = probability of option K and π = number of modes i) ii) iii) What are the probabilities each mode will be chosen? (3 marks) Due to poor use, the balloon will be eliminated during the next season. What are the new probabilities for the two remaining modes? (2 marks) Considering just the helicopter and the hovercraft again and using the original costs from the table, what value of OVTT for the hovercraft would make the probability equal? (4 marks) Solution i) (1 mark per mode) ππ -3.087 -3.362 -4.263 Helicopter Hovercraft Balloon π ππ 0.046 0.035 0.014 0.095 P 0.484 0.368 0.147 ii) Remove balloon as an option and determine the proportions (1 mark for each proportion) 0.046 Helicopter: π(1) = 0.046+0.035 = 0.568 0.035 Hovercraft: π(2) = 0.046+0.035 = 0.432 iii) Goal: π1 = π2 π1 = 0 − (0.009 × 85.40) − (0.057 × 5.35) − (0.061 × 33) π1 = −3.0866 π2 = 0.4 − (0.009 × 88.16) − (0.057 × 15.7) − (0.061 × ππππ2 ) π2 = −1.2883 − (0.061 × ππππ2 ) Equate and solve: (let Δππππ = ππππ2 − ππππ1) (2 marks) 0.061 × ππππ2 = −1.2883 + 3.0866 ππππ2 = 29.4789 (2 marks) → Assuming that ππππ1 = 33, the OVTT for the hovercraft would have to be 29.48 (1 mark) Question 6: Traffic Assignment [10 marks] a) Briefly describe the process of traffic assignment in the context of the 4-step transport model. Please present 3 objectives of the traffic assignment process and why they are important in transport appraisal. (5 marks) Solution • • Selection of routes between origins and destination of a transport network (2 marks) Objectives: (3 marks – 1 for each objective) o Estimate travel costs for each link of a road network. o To obtain aggregate network performance measures to inform planners and engineers in managing road traffic and transit operations o Gain an understanding of route utilisation and redundancies within a network. o To obtain microscopic information (turning movements and link volumes) to design the future of the network. b) Imagine you are a traffic engineer and you are faced with the following road network presented in Figure 1 below. Due to heavy congestion in the network, there is a proposal to include a new 2-way link between node B and node C. Discuss the implications of building a new links on a road network in terms of route choice as well as overall urban planning. What are the advantages and disadvantages? Are there better alternatives to mitigate congestion? (5 marks) Figure 1: Road Network for part b) Solution • • The construction of a new link between B and C creates additional routes for travelers on the network. (1 mark) Advantages: (1 mark) o Adding capacity provides more supply which can cater for the demand • • o Road construction results in employment and potentially drives economic growth. Disadvantages: (2 marks) o Induced demand o Mode shift towards private transport o Unsustainable as the induced demand may exacerbate congestion. Alternatives: (1 mark) o Invest in public and active transport (create a rail network, add a bike path)