There’s no evidence that widening the I-5 freeway at the Rose Quarter will reduce crashes.

ODOT used a model that doesn’t work for freeways with ramp-meters

When ODOT widened I-5 lanes and shoulders near Victory Boulevard in 2010, crash rates did not decline

Research shows interstate freeway shoulder widths aren’t correlated with crash rates

The marketing campaign for spending $1.45 billion to widen the I-5 Rose Quarter freeway heavily emphasizes safety–falsely claiming that this is a particularly dangerous roadway

Reducing crashes is a linchpin of the project in two ways, both to support an argument about safety, and also because the project claims that with fewer crashes they’ll be less “non-recurring delay.”  City and state officials concede that due to induced demand the project won’t do anything about daily recurring delay–any improvements in the regular flow of traffic will simply prompt more people (who now avoid the roadway because of congestion) to try to use it, with the result that it will be just as slow in future as it is before spending a billion dollars or so).

ODOT incorrectly applied the ISATe model to I-5, ignoring the effect of ramp meters.

The project claims that widening the freeway will reduce crashes.  How they know?  Well according to the project’s “Transportation Safety Technical Report,” were generated by a tool called “ISATe”.  Despite the impressive moniker, ISATe is just an excel spreadsheet.

The AASHTO HSM (AASHTO 2010) predictive method for highways and interchanges using the software ISATe was applied to estimate the relative safety performance of the proposed Project. The method was applied without calibration factors, so the results are presented as relative differences rather than absolute predictions. The models are applied on a segment-by-segment basis, and segments are defined to have consistent geometric characteristics. The largest safety benefit results from upgrading shoulders to full standard on both sides of the highway.

(Emphasis added)

This statement reveals a number of important points.  First, they didn’t calibrate the ISATe mode to existing crash rates on the roadway.  ISATe is based on a rough approximation of factors for a wide range of roadways, and to produce relative, rather than absolute values.

The report specifically says that the the methodology doesn’t account for the effects of ramp metering.  Ramp meters have already been installed on this portion of freeway,

And if you read the ISATe technical manual, there’s a section that explains the limitations of the model, helpfully labeled “Limitations of the Predictive Methods.” The manual says that the ISATe spreadsheet doesn’t estimate crashes for freeways that have ramp meters.

Bonneson, J., Pratt, M., and Geedipally, S., (et al), Enhanced Interchange Safety Analysis Tool: User Manual, National Cooperative Highway Research Program, Project 17-45, Enhanced Safety Prediction Methodology and Analysis Tool for Freeways and Interchanges, May 2012.

This is important because a key reason for ramp meters is to smooth the flow of traffic, and reduce merge conflicts that would otherwise result in an increase in crashes. What ODOT has done is use a tool that assumes that there are no ramp meters to estimate the number of crashes on a road that has long had — and will continue to have ramp meters.

Consequently, by inappropriately applying the ISATe spreadsheet to a roadway that’s already had ramp metering installed, ODOT has overstated the base level of crashes on the road, and overstated the likely reduction in crashes from wider shoulders.  According to the US Department of Transportation, ramp metering on Portland freeways has already produced a 43 percent decline in crashes.

Because the ISATe model doesn’t take account of the actual presence of ramp meters, it over-estimates the reduction in crashes that might be attributable to adding lanes and shoulders to this stretch of I-5.  In essence, the installation of ramp meters here has already accomplished some, if not most of the supposed crash reduction benefit that would be achieved by widening the freeway.

ODOT’s own experience on I-5 shows wider shoulders didn’t decrease crash rates

So rather than looking at a spreadsheet based model that, according to its own user manual, isn’t capable of accurately estimating crash rates on this kind of freeway, it makes more sense to look at the actual experience with freeway widenings on Interstate 5.  As we’ve pointed out at City Observatory, ODOT has already helpfully run this experiment. It widened a mile-long stretch of I-5 just north of the Rose Quarter in 2010, between Lombard Street and Victory Boulevard, adding lanes and widening shoulders.  That stretch of road is used by essentially the same drivers to travel through the Rose Quarter. It is, in a sense, a perfect natural experiment on the efficacy of wider freeways reducing crashes.

And what ODOT’s own data show is that after widening I-5 crash rates were higher than before the freeway widening. The data is shown below.  Crashes spiked during the construction period (2010; gray bar), which is a common occurrence.  But after project construction was completed, from 2011 onward, crashes were consistently about 10 percent higher than they had been prior to construction.  Far from reducing crashes, freeway widening actually increased them.

Research shows wider shoulders aren’t statistically associated with fewer crashes on Interstate Freeways

There’s actually no valid statistical evidence that shoulders wider than 6 feet on divided roads produce any additional safety benefit.

The same could be observed for shoulder width, where the crash rate decreases up to 6 ft and then varies as the shoulder becomes wider. These trends are simple observations, and statistical tests were not conducted to determine their statistical significance. (page 21)

Stamatiadis, N., Pigman, J., Sacksteder, J., Ruff, W., & Lord, D. (2009). NCHRP Report 633: Impact of shoulder width and median width on safety. Transportation Research Board of the National Academies, Washington, DC.

Other research has concluded that there is no statistically significant relationship between outside shoulder width and crash frequency on Interstate highways.

Highway geometric factors are found to have variable impacts on highway safety (across accident severity level and road class). For example, it is found that widening outside shoulder by 1% will have no impact on Interstate highways  . . .

Chen, S., Saeed, T. U., Alinizzi, M., Lavrenz, S., & Labi, S. (2019). Safety sensitivity to roadway characteristics: A comparison across highway classes. Accident Analysis & Prevention123, 39-50.

In short, there’s no evidence that the Rose Quarter freeway widening will improve safety:

  • ODOT used a model (ISATe) that doesn’t apply to this freeway, which has ramp meters.
  • ODOT’s own experience with widening the I-5 freeway and shoulders resulted in an increase in crash rates, not a decline
  • The scientific literature concludes that lane and shoulder width is a statistically insignificant factor in determining crash rates.