When traffic volumes are heavy enough to warrant signalization, the selection process becomes somewhat more rigorous. The usual basis for selection here is that a roundabout will provide better operational performance than a signal in terms of stops, delay, fuel consumption, and pollution emissions. For planning purposes, this may generally be assumed to be the case provided that the roundabout is operating within its capacity. The task then becomes to assess whether any roundabout configuration can be made to work satisfactorily. If not, then a signal or grade separation are remaining alternatives. As in the case of stop control, intersections with heavy left turns are especially good roundabout candidates.
The graphical approximation presented earlier for capacity estimation should be useful at this stage. The results should be considered purely as a planning level estimate, and it must be recognized that this estimate will probably change during the design phase. Users of this guide should also consult the most recent version of the Highway Capacity Manual (HCM) (10) as more U.S. data and consensus on modeling U.S. roundabout performance evolves.
As in the case of AWSC operations, some of the most important benefits of a roundabout compared to a traffic signal will accrue during the off-peak periods. The comparison of delay savings discussed previously has therefore been extended to deal with traffic signals as well as stop signs. The same temporal distribution of traffic volumes used for the roundabout-AWSC comparison was assumed.

The signal timing design was prepared for each of the conditions to accommodate traffic in the heaviest peak period. The traffic actuated controller was allowed to respond to fluctuations in demand during the rest of the day using its own logic. This strategy is consistent with common traffic engineering practice. All approaches were considered to be isolated and free of the influence of coordinated systems. Left turn protection was provided for the whole day for all approaches with a volume cross-product (i.e., the product of the left turn and opposing traffic volumes) of 60,000 or greater during the peak period. When left turn protection was provided, the left turns were also allowed to proceed on the solid green indication (i.e., protected-plus-permitted operation).

The results of this comparison are presented in Exhibit 3-12 for 50 percent major street traffic and Exhibit 3-13 for 65 percent major street traffic. Both cases include AADT values up to 34,000 vehicles per day. Single-lane approaches were used for both signals and roundabouts with AADTs below 25,000 vehicles per day. Two-lane approaches were assumed beyond that point. All signalized approaches were assumed to have left turn bays.
Benefits may continue to accrue beyond the 34,000 AADT level but the design parameters for both the signal and the roundabout are much more difficult to generalize for planning level analyses. When AADTs exceed 34,000 vehicles per day, performance evaluation should be carried out using the more detailed procedures presented in Chapter 4 of this guide.
The selection of a roundabout as an alternative to signal control will be much simpler if a single-lane roundabout is estimated to have adequate capacity. If, on the other hand, it is determined that one or more legs will require more than one entry lane, some preliminary design work beyond the normal planning level will generally be required to develop the roundabout configuration and determine the space requirements.