Questions & explanations
1. In Wardrop's user equilibrium, what condition holds for all used routes between an origin and destination?
In Wardrop's user equilibrium, all used routes between an origin and destination have the same travel time, and no driver can reduce their travel time by switching to another route. This is a stable state where each driver selfishly chooses the fastest route. The system optimum principle, by contrast, minimizes total travel time for all drivers, which may require some drivers to take longer routes. The Frank-Wolfe algorithm is a common method to solve for user equilibrium in traffic assignment. It iteratively finds a direction toward the equilibrium and moves in that direction. The method of successive averages (MSA) is another algorithm that uses decreasing step sizes to converge to the equilibrium. Dial's algorithm is a stochastic assignment method that assigns traffic probabilistically based on route travel times.
2. Explain how the Frank-Wolfe algorithm finds the user equilibrium traffic assignment. What are the main steps?
The Frank-Wolfe algorithm solves the user equilibrium traffic assignment by iteratively finding a descent direction and moving toward it. First, it initializes with an all-or-nothing assignment based on free-flow travel times. Then, it computes the current travel times and finds the shortest paths, creating an auxiliary flow pattern. Next, it performs a line search to find the optimal step size that minimizes the objective function, which is the sum of integrals of travel time functions. The algorithm updates the flow by moving from the current flow toward the auxiliary flow by the step size. This process repeats until convergence, meaning the flows change very little between iterations. It is efficient for large networks because it avoids storing the full path set.
3. Describe Dial's algorithm for stochastic traffic assignment. How does it assign traffic probabilistically?
Dial's algorithm is a stochastic traffic assignment method that assigns traffic based on the probability of choosing a route, which depends on the route's travel time. It uses a logit model where routes with shorter travel times have higher probabilities. The algorithm first computes the likelihood of each link being used, considering all possible routes. It then performs a forward pass to calculate the probability of reaching each node and a backward pass to assign flows proportionally to these probabilities. The method assumes that drivers have imperfect information about travel times. Dial's algorithm is efficient because it avoids enumerating all paths. It produces a smooth distribution of flows across the network, unlike deterministic assignment.
4. What are the advantages and limitations of the cell transmission model compared to the kinematic wave theory?
The cell transmission model (CTM) is a numerical approximation of kinematic wave theory, making it suitable for computer simulation. Its advantages include simplicity, computational efficiency, and the ability to handle complex networks and boundary conditions like traffic signals. CTM can model shockwaves and queue spillback accurately. However, it has limitations: it uses discrete cells and time steps, which can introduce numerical diffusion, smoothing sharp changes. The model assumes a triangular or trapezoidal fundamental diagram, which may not capture all traffic regimes. CTM also requires careful calibration of cell length and time step to satisfy the Courant-Friedrichs-Lewy condition for stability. Despite these, CTM is widely used in practice.
5. How does the system optimum principle differ from user equilibrium in terms of total travel time?
The system optimum principle minimizes the total travel time across all vehicles, while user equilibrium results in each driver minimizing their own travel time. In system optimum, some drivers may be assigned longer routes to reduce overall congestion, leading to a lower total travel time than user equilibrium. However, drivers may not follow system optimum voluntarily because it may increase their individual travel time. The difference between the two is called the price of anarchy. Traffic engineers use system optimum to design tolls or ramp metering to align individual incentives with system efficiency. Algorithms like Frank-Wolfe can also be adapted to find system optimum flows.
6. What is the method of successive averages (MSA) and how does it differ from Frank-Wolfe in traffic assignment?
The method of successive averages (MSA) is a stochastic approximation algorithm for traffic assignment. Unlike Frank-Wolfe, which uses a line search to find the optimal step size, MSA uses a predetermined step size that decreases with each iteration, typically 1/k where k is the iteration number. This makes MSA simpler and faster per iteration, but it may converge more slowly. MSA is often used in dynamic traffic assignment and with stochastic route choice models. It updates flows by averaging the current flow with the auxiliary flow from shortest paths. The algorithm is robust but may not converge to the exact user equilibrium if the step size does not satisfy certain conditions.
7. What is the cell transmission model (CTM) and how does it represent traffic flow?
The cell transmission model (CTM) is a discrete version of the kinematic wave theory for traffic flow. It divides a road into cells of equal length, and traffic moves between cells in discrete time steps. Each cell has a maximum number of vehicles it can hold (jam density) and a maximum flow rate (capacity). The model updates the number of vehicles in each cell based on the flow from upstream cells and the flow that can be received by downstream cells. It captures shockwaves and congestion propagation. CTM is widely used in dynamic traffic assignment and simulation because it is simple and computationally efficient. It can model traffic lights, incidents, and variable demand.
8. How does the kinematic wave theory explain the formation and dissipation of traffic jams?
Kinematic wave theory describes traffic flow as a continuum, using the conservation of vehicles and a fundamental diagram relating flow, density, and speed. Traffic jams form when the inflow exceeds the capacity of a bottleneck, causing a shockwave that moves upstream. The shockwave speed is determined by the change in flow and density. As demand decreases, the jam dissipates when the outflow exceeds inflow, creating a rarefaction wave. The theory predicts that traffic jams propagate at a constant speed relative to the road. It is the basis for many traffic models, including CTM. Understanding kinematic waves helps in designing ramp metering and variable speed limits.
9. Explain how CTM can model a bottleneck, such as a lane drop or an incident. What happens to the cells?
To model a bottleneck in CTM, the capacity of the downstream cell is reduced, e.g., from 2000 vehicles per hour to 1500. The sending flow from the upstream cell may be higher than the receiving flow of the bottleneck cell, causing the actual flow to be limited by the bottleneck's receiving flow. Vehicles accumulate in the upstream cell, increasing its density. If the upstream cell reaches its jam density, the congestion spills back to earlier cells. The model captures the queue growth and propagation. When the bottleneck is removed, the receiving flow increases, and the queue dissipates as vehicles flow out. CTM can simulate the time-space evolution of congestion.
10. Compare SCATS and SCOOT in terms of their approach to adaptive signal control.
SCATS uses a plan-selection approach, choosing from a library of pre-calculated signal plans based on current traffic conditions, while SCOOT uses a model-based incremental optimization approach. SCATS is more reactive and can handle a wide range of conditions quickly, but it requires a large set of plans. SCOOT continuously adjusts timings, making it more responsive to gradual changes, but it may be slower to react to sudden events. SCATS is more robust to detector failures because it can fall back to plans, while SCOOT relies heavily on its model. Both systems are effective, but SCATS is more common in Australia and Asia, while SCOOT is prevalent in Europe.
11. Compare ALINEA and demand-capacity algorithms: which one is more robust to changes in freeway conditions?
ALINEA is generally more robust than demand-capacity because it uses feedback control that continuously adjusts to changing conditions. Demand-capacity relies on accurate measurements of upstream demand and downstream capacity, which can vary due to incidents or weather. ALINEA only needs occupancy measurements and a set point, making it less sensitive to measurement errors. It also prevents congestion by maintaining occupancy near the critical value. Demand-capacity can cause oscillations if capacity is overestimated. However, ALINEA may require tuning of the controller parameters. Overall, ALINEA is preferred in practice for its stability and adaptability.
12. How does the demand-capacity ramp metering algorithm work, and when is it effective?
The demand-capacity algorithm measures the upstream demand (flow arriving at the merge) and downstream capacity (maximum flow the freeway can handle). The metering rate is set to the difference between capacity and the measured upstream flow, but not less than a minimum rate. If demand is low, the rate is high; if demand approaches capacity, the rate is reduced to prevent congestion. This algorithm is effective when the downstream capacity is known and stable, but it can be reactive and may cause queue spillback if not coordinated. It is simpler than ALINEA but less robust to disturbances. It works best in uncongested conditions where demand is predictable.