7. Application

This section demonstrates the application of the resulting weather-sensitive DTA model to an actual network, with particular focus on two aspects: (1) assess the impacts of adverse weather on transportation network; and (2) evaluate effectiveness of weather-related variable message signs in alleviating traffic congestion caused by adverse weather conditions.

7.1 Test bed Network and Simulation Settings

Figure 7-1 shows the test network, namely the CHART (Maryland, United States) network (Mahmassani et al., 2005). The network consists, primarily, of the I-95 corridor between Washington, DC and Baltimore, MD, and is bounded by two beltways (I-695 Baltimore Beltway to the north and I-495 Capital Beltway to the south). The network has 2182 nodes, 3387 links and 111 traffic analysis zones (TAZ). A two-hour morning peak (i.e. 7-9AM) dynamic OD demand table estimated for the network is used in the experiments. Travelers are assumed to follow their habitual routes, which are determined by performing a dynamic user equilibrium assignment, as suggested by Mahmassani and Peeta (1993). When an adverse weather event occurs, travelers will stick to their habitual routes if they do not receive specific road weather information, or are not required to detour by certain control measures. However, if such information or controls are available, for example, a weather VMS indicating extra delay on a certain road due to heavy rain, travelers might change to a better route.

This image is of the CHART highway network, which primarily consists of the I-95 corridor between Washington, DC and Baltimore, MD, and is bounded by two beltways (I-695 Baltimore Beltway to the north and I-495 Capital Beltway to the south).
Figure 7-1. The CHART Network

7.2 Network Performance under Adverse Weather Conditions

To illustrate the effects of network-wide road weather conditions, three scenarios are compared:

Scenario 1 ("Clear"): the base case scenario corresponds to clear weather conditions.
Scenario 2 ("Moderate rain"): corresponds to a moderate rain day, that is, visibility of 1.0 mile and rain intensity of 0.2 inch/hour.
Scenario 3 ("Heavy rain"): corresponds to a heavy rain day, that is, visibility of 0.5 mile and rain intensity of 0.5 inch/hour.

The time-varying network travel times are compared in Figure 7-2 for these three scenarios. Since the rainy weather affects the supply-side parameters, such as lower capacity and saturation flow rate, the network travel times become longer when the weather conditions get more serious.

Graph shows average travel time and departure times during clear, moderate rain, and heavy rain conditions. Each curve starts out at about 12 minutes of travel time at 7:00AM and increases as departure time increases. Clear increases to about 15 minutes of travel time, moderate rain increases to about 25 minutes of travel time, and heavy rain increases to about 35 minutes of travel time.
Figure 7-2 Network Travel Time Comparison

A similar pattern, namely heavier rain condition resulting in longer travel time, is obtained by examining time-varying travel times between a major OD (origin-destination) pair of the network, as shown in Figure 7-3. Moreover, Figure 7-4 shows the standard deviations of actual travel times at 5-minute intervals. We can see that not only travel time becomes longer when adverse weather occurs, but also the variability of the travel time is greater, making travel less reliable in the network.

Figure 7-3. OD Travel Time Comparison

Bar chart shows standard deviation and departure times in clear, moderate rain, and heavy rain conditions. As departure time increases, the standard deviations of each increase. At 7:00AM departure time, all three bars have a standard deviation beneath 2 minutes. At 9:00AM departure time, clear and moderate rain has a standard deviation time of around 8 minutes while heavy rain has a standard deviation of around 11 minutes.

Figure 7-4. Standard Deviations of OD travel times

In addition, to examine the impacts of local weather condition (for instance, rain in a certain area of the network) we assume that it rains on a stretch of Freeway I-95, following the pattern shown in Figure 7-5. Namely, the rain starts at 7:10 AM with visibility of 1.0 mile and intensity of 0.2 inch/hour; at 7:40 AM the rain intensity increases to 0.5 inch/hour and the visibility decreases to 0.5 mile, indicating heavier rain situation; then at 8 AM the rain stops.

Chart combines time, rain intensity, and visibility conditions. The graph shows that rain intensity and visibility correlate: when rain intensity is at zero inches per hour, visibility is at 10 miles. Any increase in rain intensity greatly decreases visibility. When rain intensity is 0.2 inches per hour or 0.5 inches per hour, the visibility is approximately 0.05 miles.
Figure 7-5. Time-Varying Local Weather Conditions

A weather-impacted link along I-95 is selected to illustrate local weather impacts on traffic conditions. The speed on the link maintains (nearly) the free flow speed when the weather is clear. With precipitation, however, the speed drops significantly. For the 7:10-7:40 AM time period, when it rains moderately, the speed drops to around 50 mph. This is mainly caused by drivers' speed reduction in response to low visibility and slippery road surface, which could be predicted/calculated using weather adjustment factor shown in Equation (5-1). Nevertheless, in the 7:40-8:00 AM period, when it is raining heavily, the speed drops to as low as 20 mph. This is a combined outcome of drivers' speed reduction response and congestion effects; that is, drivers slow down because of not only the precipitation conditions but also the traffic congestion along the road. Under such circumstances, predicting link performance using the WAF alone, while ignoring the congestion effect caused by the adverse weather conditions, would not be sufficient.

Chart shows time and speed for clear conditions and rain. During rain conditions, the speed is always lower than during clear conditions. The clear condition line stays fairly constant at around 65 miles per hour, while the rain condition line ranges from 65 miles per hour to a low of 20 miles per hour.
Figure 7-6. Time-Varying Speeds on a Weather-Impacted Link

7.3 Evaluation of Weather-related Information and Control Strategies

In order to alleviate the impacts of adverse weather, weather-related information and control strategies could be applied. Three scenarios are compared to illustrate the effectiveness of weather detour VMS.

Scenario 1 ("Clear"): the base case scenario corresponds to clear weather conditions. Users are assumed to follow their habitual routes, that is, the user equilibrium.

Scenario 2 ("Rain"): the time-varying weather conditions (Figure 7-6) are applied to a stretch of I-95. Travelers, however, still follow their habitual routes, as there is no road weather information or control measures.

Scenario 3 ("Rain + VMS"): variable message signs are placed upstream of the weather-impacted area, which indicate an extra delay (penalty) on each impacted link due to adverse weather conditions. In particular, the penalty is specified as 50% of link travel time for the moderate rain period (i.e. 7:10-7:40AM) and 100% for the heavy rain period (i.e. 7:40-8:00AM).

Figure 7-7 to Figure 7-9 present the link performance under these three scenarios. Weather VMS is able to detour some travelers from the impacted road, reflected in less accumulated flow (as shown in Figure 7-9) on the rain-affected link. As a result, the congestion effect caused by the precipitation condition is eliminated. As shown in Figure 7-7, the speed reductions during the moderate rain period (7:10-7:40AM) are comparable whether VMS is active or not. As explained earlier, this reflects drivers' voluntarily speed reduction to accommodate worse driving conditions for safety concerns. During the heavy rain period, however, weather VMS helps to maintain relatively high speed and relatively low density compared to the no-VMS case. The congestion caused by adverse weather is clearly shown by link density comparison (Figure 7-8). Under clear condition or using weather VMS, the density is kept below 30 vehicles per mile per lane (LOS D or better), indicating relatively uncongested traffic conditions. On the other hand, when there is precipitation on the road but travelers are not informed, heavier congestion is experienced.

In brief, weather detour VMS helps to alleviate traffic congestion by detouring travelers to alternative routes. The voluntary speed reduction due to safety concern is, however, not affected by this type of VMS.

Chart shows time and speed for clear, rainy, and rainy conditions plus VMS.
Figure 7-7. Link Speed Comparison

Graph shows time and density for the base condition, the rain condition, and the rain plus VMS condition.
Figure 7-8. Link Density Comparison

Graph shows time and accumulated flow for the base condition, the rain condition, and the rain plus VMS condition.
Figure 7-9. Accumulated Flow Comparison

In addition, the time-varying travel times of vehicles traveling between an OD pair (affected by the rainfall area) are compared. As shown in Figure 7-10, the precipitation, especially the heavy rain period, affects driving conditions and therefore results in longer travel time. Variable speed limits however help maintain relatively low travel time as the OD flows are distributed more efficiently in response to the speed limit indication. This observation is consistent with Abdel-Aty et al.'s (2006) conclusion that VSL strategies, when properly set, could produce travel time savings, in addition to their potential safety benefit.

This graph shows the OD travel time comparison flow for the clear condition, the rain condition, and the rain plus VSL condition.
Figure 7-10. OD Travel Time Comparison

Previous | Next