Getting Trip Estimates Right at Signalized Intersections
Trip generation at a signalized intersection isn't just about plugging numbers into a chart and moving on. The Highway Capacity Manual has the methodology, sure, but the real work happens when the model output doesn't match what's actually turning left off a new strip mall at 5:15 PM on a Tuesday. The basic idea is straightforward. You estimate how many vehicles arrive at an intersection from each approach based on the surrounding land uses, then distribute those vehicles across the available movements. Trip generation feeds directly into your capacity analysis, signal timing design, and sometimes your impact fee calculations. Get it wrong and your level of service estimates will be off by enough that the city engineer will send the whole thing back. I used to rely heavily on the ITE Trip Generation Manual tables. They're convenient. They're also often wrong for your specific site because the data behind them is aggregated across dozens of different climates, regions, and development types. A suburban office park in Florida generates traffic differently than a similar one in Minnesota, but the base tables treat them almost identically.
Here's what most people skip: the conversion from annual average daily traffic to the design hour volume. The manual gives you a trip rate per unit, but you need to figure out which hour matters. Peak hour factor, distribution factor, and turn ratio all come into play. A common mistake is applying the 30th highest hourly volume factor to a site that actually peaks differently due to nearby school zones or hospital shift changes. I ran into this on a project outside Columbus a few years back. The ITE tables suggested a moderate trip rate for a medical office development, but the actual peak came from a combination of the office workers and the adjacent hospital's evening shift change around 6:30 PM. That meant the peak hour wasn't the traditional 5 to 6 PM window at all. I ended up pulling actual turn movement counts from two nearby signalized intersections with similar land uses, adjusted for site-specific variables, and used those calibrated rates instead of the handbook defaults. The difference was roughly 40 percent more right-turn volume than the tables predicted. For the distribution part, you need to think about where those trips are going. If an intersection has three major generators on the north approach and nothing on the south, you aren't getting balanced turning movements. I always map the trip attraction points within a half-mile radius and weight them by distance and road hierarchy. The default five-split assumption in some software packages is fine for rough screening but falls apart quickly in the field.
One thing that trips people up is the treatment of external trips. Through traffic that doesn't originate or terminate at the study area still passes through your intersection and eats into capacity. If you're doing a trip generation analysis for a new development and ignoring the external through flow, your delay estimates will be unrealistically low. I make it a habit to pull count data from the nearest upstream and downstream signals and back-calculate the through component. Takes about ten minutes if you have the data access. Another nuance worth noting: pedestrian and bicycle trips get folded into the volume in some analyses but not others, depending on what your agency requires. In dense urban cores, a bike share station half a block from an intersection can add meaningful turning movements that aren't captured by auto-only trip rates. I once saw a downtown signal timing plan fail because nobody accounted for the e-bike delivery traffic surge during lunch hours. The intersection ran red during the actual peak, not because of cars, but because of a hundred extra bike trips in a single cycle. If you want better accuracy, consider pairing your trip generation with a simple gravity model for destination assignment rather than relying on fixed distribution percentages. It takes maybe twenty minutes more per intersection and usually pays for itself in fewer redesigns.
Get the Full Details
The biggest limitation of the standard trip generation approach is that it assumes stable land use patterns. When a major employer opens or closes, or when a new transit line starts running nearby, the historical trip rates become obsolete quickly. In those cases, the only reliable approach is primary data collection. Count the intersections yourself for at least a full week, covering both weekdays and weekends. It costs more upfront but saves you from fixing a bad signal timing plan six months later.
