Summary and conclusion

In transportation simulation models for larger scale questions such as planning, the flow characteristics of the traffic dynamics are in some sense more important than the microscopic driving dynamics of the vehicles itself. This becomes especially true since a ``complete'' representation of human driving is impossible anyway, both due to knowledge constraints and due to computational constraints. Yet, calibrating a traffic simulation model against all types of desired behavior (for example against all HCM curves and values mentioned in this paper) seems a hopeless task given the high degrees of freedom.

Transims thus attempts to generate plausible emergent macroscopic behavior from simplified microscopic rules. This paper described the more important aspects of these rules as currently implemented or under implementation in TransimsBefore we implement rules in the Transims production version, we usually try to run systematic studies with more experimental versions. The results of the traffic flow behavior from that study were presented. Also, we showed the effects of some changes in the rules for the example of a yield sign. Finally, some comparisons were made between the logic currently under implementation and the logic used for the Dallas/Fort Worth case study.

One problem with microscopic approaches is that, in spite of all diligence, subtle differences between design and actual implementation can make a significant difference in the emergent outcome. For that reason, this paper should also be seen as an argument for a standardized traffic flow test suite for simulation models. We propose that simulation models, when used for studies, should first run these tests to demonstrate the dynamics of their emergent macroscopic flow behavior. We think that the combination of results presented in Figs. 32.2 to 32.5 are a good test set, although extensions may be necessary in the future (e.g. merge lanes, weaving, etc.). We will attempt to provide future Transims results also with updated versions of the results of the traffic flow tests.

Figure: (a) Definition of $gap$ and examples for one-lane update rules. Traffic is moving to the right. The leftmost vehicle accelerates to velocity 2 with probability 0.8 and stays at velocity 1 with probability 0.2. The middle vehicle slows down to velocity 1 with probability 0.8 and to velocity 0 with probability 0.2. The right most vehicle accelerates to velocity 3 with probability 0.8 and stays at velocity 2 with probability 0.2. Velocities are in ``cells per time step''. All vehicles are moved according to their velocities at a later phase of the update. (b) Illustration of lane changing rules. Traffic is moving to the right; only lane changes to the left are considered. Situation I: The leftmost vehicle on the bottom lane will change to the left because (i) the forward gap on its own lane, 1, is smaller than its velocity, 3; (ii) the forward gap in the other lane, 10, is larger than the gap on its own lane, 1; (iii) the forward gap in the target lane is large enough: $weight2 = v - gap_o = 3 - 10 = -7 < 1 = weight1$; (iv) the backward gap is large enough: $weight3 = v_{max} - gap_b = 5 - 6 = -1 < 1 = weight1$. Situation II: The second vehicle from the right on the right lane will not accept a lane change because the gap backwards on the target lane is not sufficient. (c) Value of $weight4$ when in wrong lane during the approach to the intersection. (d) Example of a left turn against oncoming traffic. The turn is accepted because on all three oncoming lanes, the gap is larger or equal to three times the first oncoming vehicle's velocity. (e) Test networks.

(a) (b)

(c) (d)

(e)

Figure: One-lane traffic: Flow vs. density, travel velocity vs. flow, and travel velocity vs. density.

Figure: Three-lane circle: Flow vs. density, travel velocity vs. flow, travel velocity vs. density, lane usage vs. flow, and land usage vs. density. The asymmetry in the lane usage at low densities is due to the fact that the parking locations start filling in vehicles on the right lane, and they only move to the left when traffic on the right lane becomes dense.

Figure: of vehicles going through the intersection and number of vehicles ``off plan'' ($=0$) per green phase, re-scaled to hourly flow rates per lane.

Figure: through stop sign, yield sign, and unprotected left turn. Left column: Major road (``circle'') has one lane. Right column: Major road (``circle'') has two lanes. Solid line: Highway Capacity Manual (114). $v_{max}=3$, gap acceptance rule is ``accept if $gap \ge 3 \cdot v_{oncoming}$, and if first site on target lane available''. Note that for ``left turn across two lanes'' (bottom right) the opposing volume is the sum of both lanes, i.e. twice the value shown on the x-axis.

Figure: Comparison between different rules for the case of a 1-lane minor road controlled by a yield sign merging into a 1-lane major road. (a) Same as Fig. 32.5 (i.e. $v_{max}=3$ and ``accept if $gap \ge 3 \cdot v_{oncoming}$''), except that traffic on major road does not reserve the first cell on the outgoing link, thus giving traffic from the yield sign more opportunities. Note that this seemingly small difference has big consequences in the congested regime. (b) Same as (a) except that acceptance rule now ``accept if $gap > 3 \cdot v_{oncoming}$''. (c) Same as (b) except that $v_{max}=5$. (d) Same as (b) except that $v_{max}=2$. (e) Same as (b) except that acceptance rule now ``accept if $gap > v_{oncoming}$. (f) Same as (b) except that acceptance rule now ``accept if $gap > v_{max}$''.

Figure: between the March 1998 Transims microsimulation gap acceptance logic and the one used in the case study. Flow through stop sign, yield sign, and unprotected left turn into/across traffic on major road. Left column: March 1998 Transims microsimulation. Right column: case study Transims microsimulation. The arrows in the left turn case indicate the direction of increasing congestion. - The results are not strictly comparable because (i) the simulations in the right column were run with a maximum speed of $v_{max}=5$ cells/update (135 km/h) vs. $v_{max}=3$ cells/update (81 km/h) in the left column (mostly noticeable in the lower maximum flow on the major road); and (ii) the stop and yield cases on the right describe flow into a 3-lane road vs. flow into a 1-lane raod in the left column. Note that the results for the turns into other traffic (``stop'' and ``yield'') are not that much different between the two whereas the result for the turns across other traffic (``left turn'') leads to much higher flows in the uncongested and lower flow in the congested regime with the case study logic.