+1
- 1. +1 Jun Su Park After completing the research for my research question, I have determined that the ideal signal-timing method that should be employed at oversaturated intersections should be the algorithm based on maximizing throughput. I have examined each prevalent method used at oversaturated intersections (fixed, delay minimizing, maintaining queue ratio) and came to this conclusion. To understand the logic behind this decision, I will go through each method that was covered and summarize its benefits and drawbacks. The first method is the fixed- timing method. This strategy was not mentioned as it cannot be described as a strategy—it is simply a signal with each approach having constant green times for every cycle. This can be described as the most basic method at intersections to control traffic and they simply serve the purpose to stop cars from colliding into one another rather than to reduce congestion. Therefore, the fixed-timing method is strongly not recommended for oversaturated intersections, as it does not alleviate the situation and can even make it worse by allocating an excess of green time to unnecessary approaches. The second method that was discussed was the discrete minimal delay model and the performance index model. Both are new versions of previous delay-minimizing strategies and they aim to find the optimal switch-over time. The discrete minimal delay method serves to simplify delay estimation techniques used in previous research that are used to determine switch-over times and the performance index model builds on the discrete model by adding a factor of vehicle stopping times. The performance index model is more accurate due to the added factor. This method provides a significant advantage over fixed-signal timing methods as it determines the amount of green time that should be given to each approach over the whole oversaturation period while considering delay times at each approach. The third strategy that was covered was the vehicle actuation (VA) strategy. This method utilizes sensors at intersections to determine when and how much green time to give to each approach in an intersection. This source evaluated a more efficient VA strategy, called enhanced VA strategy and found that it resulted in a 30% decrease in delay compared to conventional VA strategies. Similar to the previous method, VA strategies also aim to minimize delay times. However, when VA strategies are used in oversaturated situations where the whole queue cannot disperse in one cycle with maximum green time, the method is no different from the fixed-signal method, as the sensors will continue to detect vehicles and give maximum green time each cycle. Another drawback is the cost of the sensors, which my cancel out the positive congestion relief effects. Therefore, VAstrategies are more appropriate in intersections that must be kept clear for the whole network to function, but do not have heavy traffic. The fourth method that was analyzed was the discrete-event max-plus model, which aims to minimize queue length. Just like the two previous methods, this algorithm also produces optimal signal times for each approach in the intersection when applied, but minimizes queue length instead of delay times. The authors claim that this method is more efficient, as using queue length data from sensors are much more accurate and quicker than estimating delay times. The data from the final source seem to support this claim, as it indicates
- 2. that queue minimizing models fared slightly better than delay minimizing models. One drawback is that the sensors may not be appropriate for smaller intersections to be worth the money, similar to the VA method sensors. The last method was the throughput maximizing algorithm. This algorithm aims to produce ideal cycle times for each approach, which is similar to the previous methods that were discussed. The authors assert that throughput maximization is the ideal optimization category, as previous research assume that saturation flow is constant. Also, delay minimization could be inefficient as when queues get longer, delay gets infinitely bigger and throughput maximization naturally leads to delay minimization. Another advantage is that this method requires less data such as queue length and demand of intersection usage, which makes the method more exact than previous methods. However, a drawback is that queue length have to be sufficiently long enough for this method to work. Combining all of the benefits and drawbacks, we can come to the conclusion that throughput maximization is the most efficient method compared to the other methods. Queue minimization and VA control methods require sensors, which improve accuracy but result in high costs, which may not be appropriate for isolated intersections. Delay minimization techniques are a more suitable choice that rivals throughput maximization techniques, but since most delay minimization techniques’assumptions are based on that saturation flow is constant, this may rather increase delay, as the author of the last method suggests. Also, since the delay has to be estimated, it affects the method’s overall accuracy. Applying this to the practical problem, I have come to the conclusion that the current fixed-time strategy at the four-way intersection should be changed to the throughput maximization strategy. Since this intersection is not a part of any major road network such as intersections in the city center, sensors will incur more costs than potential benefits, which eliminates the two strategies proposed. If the queue on the intersection was shorter, delay minimization techniques may be viable, but since the queue during peak times require more than 20 cycles to resolve, delay times are too great to be analyzed through this mechanism. The conditions for the throughput maximization technique are met, as the queues are long enough that they are not dispersed in a single cycle. This method also takes account that saturation flows are not constant and uses less variables that need estimation. Therefore, this method, an effective and cheap strategy, should be employed in the intersection.