CVEN 4402 #2Statistics

Java Python CVEN 4402

Assignment #2

Problem 1: Deterministic User Equilibrium (15 points).

Consider the network below with the following link travel time functions (xa  is the flow on link a and ta (xa ) is the travel time of link a):

t1 (x1) =  x1  + 2

t2 (x2) =  2x2  + 1

t3 (x3) =  3x3  + 1

t4 (x4)  =  19

t5 (x5) =  x5  + 1

Let D be the demand for the O-D pair (1, 4).

Questions:

Part A (2 points). State the flow conservations equations and, assuming all paths are used, state the User-Equilibrium equations.

Part B (3 points). Assume now that the network is empty, that is: x1   = x2   = x3   = x4  = x5  = 0. Which path will the 1st user that wants to go from 1 to 4 take? Which path will the 2nd user and 3rd user take (assuming that the users before them remain on the network)? Justify all your answers in a single sentence.

Part C (4 points). Assuming all paths are used at User-Equilibrium, express the link flow x5 as a function of D (show all work and simplify the resulting equation).

Part D (3 points). Assuming the function obtained at Part D holds, find the link flows x1,x2, x3, x4  and x5  for D  = 20 (show all work). In this context, which paths are used at User-Equilibrium?

Part E (3 points). Assuming the function obtained at Part D holds, find the link flows x1,x2, x3, x4  and x5  for D  = 40 (show all work). In this context, which paths are used at User-Equilibrium?

Problem 2: MSA and Frank Wolfe for Deterministic UE (15 pts)

Consider the network defined by the node-link incidence matrix below:

and the following link travel time functions:

t1 (x1)  =  2x1  + 3

t2 (x2) =  x2  + 4

t3 (x3) =  x3  + 2

t4 (x4)  =  3x4 +1

t5 (x5) =  2x5  + 1

t6 (x6) = 2x6  + 1

There are two origin-destination pairs (1,4) and (2,5) with demands (r is the origin node, s is the destination node and Qrs  is the travel demand from r to s):

Q14  = 40

Q25  = 30

Part A (2 points). Draw the network defined by E.

Part B (3 pts). State the objective function for the user equilibrium problem and expand it using the cost functions for the network shown above (i.e., integrate each cost function  and simplify).

Part C (5 pts): For this network, solve the UE problem using MSA. Do only 3 iterations.

Part D (5 pts). Use the results from the 3rd  iteration from Part B (i.e., x3) as the starting solution and do one iteration of the Frank Wolfe Method.

Problem 3: System Optimal Assignment (25 pts)

Consider the network from Problem 2 but with different travel time functions and demands (xa is the flow on link a and ta (xa ) is the travel time of link a):

t1 (x1) = 2x1  + 2

t2 (x2) = x2  + 5

t3 (x3) = x3  + 1

t4 (x4) = 2x4  + 1

t5 (x5) = x5  + 1

t6 (x6) = x6  + 2

There are two origin-destination pairs (1,4) and (2,5) with demands (r is the origin node, s is the destination node and Qrs  is the travel demand from r to s):

Q14  = 40

Q25  = 20

Part A (10 pts) Using the following definitions of link marginal cost, path marginal cost and System-Optimal (SO) conditions and assuming that all paths are used, find the link flows at SO conditions. State the path marginal costs of all used paths at SO conditions. Show all work.

Link Marginal Cost: Let ta (xa ) be the travel time < CVEN 4402 Assignment #2Statistics i>function of link a. The marginal cost of

link a is ta(xa) + xa dxa/dta(xa)

Path Marginal Cost: The path marginal cost is the sum of the link marginal costs on all links on this path.

SO conditions: At SO conditions, all used paths have an equal path marginal cost.

Part B (10 pts) Using the SO flow pattern found in Part A, compute the marginal cost tolls for each link. Assume the VOT, α = $2/ unit time.

Part C (5 pts) The final User Equilibrium (UE) link flows (without marginal cost tolling) are shown below. Compare the path travel times between the UE and SO assignment patterns. What is the (%) savings in total system travel time (TSTT) achieved with the SO flow pattern compared with the UE flow pattern?

Final UE link flows:

Link 1:   21

Link 2:   19

Link 3:   9.5

Link 4:   10.5

Link 5:   19

Link 6:    9.5

Problem 4: DUE with Elastic Demand (15 pts)

Consider the network below with the following link travel time functions (xa  is the flow on link a and ta (xa ) is the travel time of link a):

t1 (x1) = 2x1  + 1

t2 (x2) = x2  + 2

t3 (x3) = 3x3  + 1

t4 (x4) = 3x4  + 1

t5 (x5) = x5  + 1

t6 (x6) = x6  + 2

Assume the demand function for each OD pair A-C; A-D; B-C; B-D is:

Drs (krs )  =  20 − krs /10

Solve the UE problem with Elastic Demand using MSA. Do only 4 iterations.

You can use Excel to solve the problem.

Problem 5: SUE (30 pts)

Consider the following network:

As presented above, consider the following cost functions for the links of the network:

t1 (x1) =  4  +  2x1

t2 (x2) =  6  +  6x2

t3 (x3) =  3

t4 (x4) =  3  +  2 x4

t5 (x5) =  2  +  1x5

t6 (x6) =  3  +  4x6

Assume that a demand of 14 units travel from node 1 to node 4.

Part A (5pts).  Using the MSA approach, present the first two iterations of a Stochastic User Equilibrium traffic assignment solution by hand, assuming a logit model with a perception error value of θ  = 0.1.

Part B (5pts). Determine the converged SUE equilibrium solution for θ  = 0.1. Let E  =

1% be the convergence tolerance where the following expression must hold for all links of the network in order to achieve convergence (i.e. stop when change in each link flow is below 1% between iterations).

You can use a spreadsheet to solve this problem.

Part C (20 pts) Determine the converged SUE equilibrium solution for θ  = 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9 and answer the following questions (Please use a spreadsheet to solve, and submit only the final converged solutions for all θ in a single table):

i. (10 pts)What are the values of the converged link flows for each theta

value and how many iterations were required to achieve convergence for each case?

ii. (5 pts) Plot the variation in path costs against theta (refer to plot below). At what theta values do you observe a change in “path ranking”? (Path ranking is the order of paths from least cost to highest cost).

iii.      Describe how the value of theta affects the equilibrium solution