Table 16.1. Final segmentation for Oregon US-20 study route.

The vertical alignment along this route varies considerably. Consequently, Figure 16.7 only shows the grades that meet the segmentation criterion of vertical class 2 or higher.

Regarding the vertical alignment, the following decisions were made:

• Grades 1 and 2, in Segment 3, were similar and thus combined into a single grade, which was of Class 3.
• There is a grade that starts in the second half of the segment preceding the second passing lane segment (#27). This grade is not considered in Segment 26. The grade levels off in the last fourth of the passing lane segment. The passing lane is not split based on grade and instead, Grade 4 is applied to the full length of the passing lane.

Segmentation Through the Town of Sisters, Oregon

The stretch of roadway through the town of Sisters consists of the following segment types (starting in the northwest portion of the town):

• One lane.
• 2,177 ft long.
• First half is undivided, marked as passing prohibited in oncoming lane; second half is divided with a median.
• Shoulder for bicycle lane, no curb or gutter.
• Posted speed limit is 35 mi/h.

This segment is analyzed as a two-lane highway passing constrained segment.

• One lane.
• 1,495 ft long.
• Undivided, marked as passing prohibited in oncoming lane.
• Shoulder for bicycle lane, no curb or gutter.
• Posted speed limit is 35 mi/h.

This segment is analyzed as a two-lane highway passing constrained segment.

Segments 1–3 are shown in Figure 16.8.

• One lane.
• Approximately 4,000 ft long.
• Undivided.

• No passing allowed in the oncoming direction.
• Several TWSC intersections.
• Some portions include a bicycle lane.
• Some portions include on-street parallel parking.
• Some portions include curb and gutter.
• Posted speed limit is 20 mi/h.

This section of roadway is analyzed as an “urban street” segment. Additional details on the segments within the urban street (hereafter referred to as “subsegments”) are as follows:

• Subsegment 1: from the point where 20 mi/h posted speed limit starts to Pine St:
  • Subsegment 1 includes bicycle lane.
  • The eastbound approach to Pine St includes a left-turn bay.

• Subsegment 2: from Pine St to Larch St:
  • Subsegment includes curb and gutter.
  • This stretch of roadway includes 100% curbside parallel parking.
  • It includes the TWSC intersections of Oak St, Elm St, Fir St, and Spruce St.
  • The eastbound approach to Larch St does not include a left-turn bay.
• Subsegment 3: from Larch St to Locust St:
  • Subsegment includes curb and gutter.
  • It includes bicycle lane.
  • The eastbound approach to Locust St includes a left-turn bay.
• Subsegment 4: from Locust St to the point where the 20 mi/h posted speed limit ends (just southeast of E Jefferson Ave):
  • Subsegment 4 includes bicycle lane.

These respective subsegments are shown in Figures 16.9 through 16.12.

• One lane.
• 1,460 ft long.
• Undivided, marked as passing prohibited in oncoming lane.
• Paved shoulder, no bicycle lane.
• No curb and gutter.
• Posted speed limit is 35 mi/h.

This segment is analyzed as a two-lane highway passing constrained segment.

The final segmentation is shown in Table 16.2. The total length is 25.66 mi.

Influence Area and Adjusted Length Calculations

The upstream and downstream influence area must be calculated for the roundabout intersection. Based on these values, adjusted segment lengths must be calculated for the roundabout segment and the upstream/downstream segments connected to the roundabout.

Roundabout (Segment #9)

Upstream influence area, per Equation (2.3)

402.15 + 10.21 × AprAvgSpeed − 15.27 × AvgCircSpeedMiHr

402.15 + 10.21 × 37.7 − 15.27 × 15 = 558.0 ft

Downstream influence area, per Equation (2.4)

− 313.80 + 32.73 × ExitSpeedMiHr − 27.01 × CirculatingSpeedMiHr

− 313.80 + 32.73 × 37.54 − 27.01 × 15 = 515.0 ft

Adjusted Segment Length: 558.0 + 515.0 = 1,073.0 ft (0.2032 mi)

The input upstream and downstream distances (440 ft, 357 ft) affect the adjusted lengths of the upstream and downstream connecting segments, as follows:

Upstream Segment (#8)

Input length: 2,177 ft (0.4123 mi); Adj. length: 2,177 + (440 − 558) = 2,059 ft (0.3900 mi)

Downstream Segment (#10)

Input length: 1,495 ft (0.2831 mi); Adj. length: 1,495 + (357 − 515.0) = 1,337.0 ft (0.2532 mi)

The urban street segment is not bounded by intersections that require the major through movement to STOP or YIELD; thus, influence area calculations are unnecessary for this segment.

Traffic Data

Traffic data were obtained from the Oregon DOT Transportation Data Management System (https://www.oregon.gov/odot/Data/Pages/Traffic-Counting.aspx).

There are 10 sensor locations along the route limits for the LOS analysis that are currently used or have been used to collect data within the last several years. These locations are summarized in Table 16.2. Note that sensors are listed in order of westernmost to easternmost location along the route.

The first milepost in the study area is milepost 91. The route continues through milepost 99, which is the last milepost before reaching the roundabout and downtown Sisters. The milepost then resets to 0 at the intersection of US-20 and SR-126, just SE of Sisters. The end of the study area is just after milepost 14.

Table 16.3 summarizes the most recently available values for AADT, percentage of traffic volume occurring in the peak hour of the day (K), and percentage of traffic volume traveling in the peak direction of the peak hour (D). The values shown in this table are for the most recent years for which actual field measurements were taken, not values estimated from growth

Table 16.3. US-20 sensor data—K, D, AADT.

projections. The K values for some of the sensors are unusually high (> 15%); however, they are not used for any subsequent volume calculations in this analysis.

The PM peak hour volumes generally occurred between the hours of 4:00 and 6:00 PM, with the highest hour typically being 4:00 to 5:00 PM. Again, the values shown are generally based on the most recent field measurements. If multiple measurements were taken during a given year, common dates across detectors were selected if possible. For example, if one sensor had counts taken during May and September of a given year and another sensor had counts taken during just May of the same year, the counts for May would be selected for both sensors. Furthermore, if multiple days of data were available in the same month of the same year, the same day of the week would be chosen across the sensors if possible.

The PHF values range from approximately 0.84 to 0.92. A single PHF value is used because specific traffic peaking times will likely vary over a route of this length as well as not to over-complicate the process of “conserving” vehicles throughout the full length of the route when setting traffic demand values. For this analysis, an approximate mid-range value of 0.89 is used. Consequently, this value effectively increases the demand flow rate for analysis purposes by 12%. The original PM peak hour volumes and corresponding values as adjusted by the PHF (rounded to the nearest 50 vehicles) are shown in Table 16.4. The PHF values are then set to a value of 1.0 in the input data settings.

For some of the sensor counts, only two-way values were provided. In these cases, the directional distribution value from Table 16.3 was used to calculate the analysis and opposing direction traffic volumes. The only volume counts available for Sensor 45996 were for the Labor Day period in 2019 (Thur 8/29, Fri 8/30, Sun 9/1, Mon 9/2). These daily volumes were almost double the average daily volumes. Thus, for this analysis example, these sensor volumes were not considered and are omitted from the table.

With a total of nine sensor locations spatially distributed across 25 miles of highway, determining locations to affect volume changes along the route is a very approximate process. To inform this process, satellite photography of the route and its surrounding area was reviewed. More major intersecting roadways—indicated by number of lanes, turning movement accommodation from the major roadway, and/or density of land use accessed by intersecting road within immediate area—were typically chosen as the locations to implement the volume changes.

The assignment of volumes (rounded to the nearest 50 vehicles) to segments, and locations where volume changes are implemented are as follows:

• Sensor 1510 is located within Segment 1. Its volume of 350 veh/h is applied to Segments 1–3. Opposing direction volume is 250 veh/h.
• Sensor 09014 is located within Segment 4. There is a net volume increase of 50 veh/h at this sensor. This net volume change of +50 is implemented by assuming a net volume change of +50 veh/h at the intersection with Hawks Beard Rd (start of Segment 4, access to Black Butte Ranch resort area). Its volume of 400 veh/h is applied to Segments 4–9. Opposing direction volume is 250 veh/h.

• Sensor 1512 is located within Segment 10. There is a net volume increase of 200 veh/h at this sensor. This net volume change of +200 is implemented by assuming a net volume change of +200 veh/h at the roundabout intersection with W Barclay Dr (Segment 9). Some of this net volume change may also occur at the intersection with Railway Rd, upstream of the roundabout, but implementing a segment break at this intersection would create a very short segment between Railway Rd and the roundabout.
• Sensor 1445 is located within Segment 10. Sensors 1512 and 1445 are located within the same segment. There is a net volume change of −50 veh/h between these two detectors, which likely occurs at the intersection with W Hood Ave. However, as was the case upstream of the roundabout, implementing a segment break at this intersection would create a very short segment between the roundabout and W Hood Ave. The higher volume for Sensor 1512, the more conservative value is retained for this segment.
  • Analysis direction volume in Segment 10 is 600 veh/h. Opposing direction volume is 400 veh/h.
• Sensors 1446 is located within Segment 11 (arterial). There is no net change in the analysis direction volume at this sensor. There is a change in the opposing direction volume; however, this volume is not used in the arterial analysis.
  • Analysis direction volume in Segment 11 is 600 veh/h.
• Sensor 1447 is also located within Segment 11 (arterial). There is a net volume increase of 150 veh/h at this sensor. This net volume change of +150 is implemented by assuming a net volume change of +50 veh/h at the intersection with Locust St (start of Subsegment 4 within the arterial segment).
• Sensor 1513 is located within Segment 13. There is a net volume decrease of 150 veh/h at this sensor. This net volume change of −150 is implemented by assuming a net volume change of
  • 150 veh/h at the intersection with SR-126/McKenzie Hwy. This intersection is located shortly downstream of where Segment 12 starts, and to avoid very short segments, this volume change is implemented at the start of Segment 12. While this traffic volume is based on a count from 2021 (2019 values were not available), the analysis and opposing direction volumes are very similar to those calculated with the 2019 AADT, K, and D values.
  • Analysis direction volume in Segment 12 is 600 veh/h. Opposing direction volume is 500 veh/h.
  • Analysis direction volume in Segment 13 is 600 veh/h. Opposing direction volume is 500 veh/h.
• Sensor 09015 is located within Segment 26. There is no net change in the analysis or opposing direction volumes at this sensor. All the intersecting roads along US-20 from Segments 14–26 are very minor.
  • Analysis direction volume in Segments 14–26 is 600 veh/h. Opposing direction volume in Segments 14–26 is 500 veh/h.
• Sensor 1515 is located within Segment 33. There is a net volume increase of 300 veh/h at this sensor. This net volume change of +300 is implemented by assuming a net volume change of +300 veh/h at the intersection with 7th St in Tumalo. While this traffic volume is based on a count from 2021 (2019 values were not available), the analysis and opposing direction volumes are very similar to those calculated with the 2019 AADT, K, and D values.
  • Analysis direction volume in Segments 27–32 is 600 veh/h. Opposing direction volume in Segments 27–32 is 500 veh/h.
  • Analysis direction volume in Segment 33 is 900 veh/h. Opposing direction volume is 750 veh/h.

Table 16.5. US-20 sensor data—PM peak hour truck percentages.

Vehicle classification data are available from only two sensor locations (09014 and 09015) and for a very limited number of dates. The truck percentages for the hour of 4:00 to 5:00 PM for the southeast-bound direction, rounded to the nearest integer value, and the segments to which they are applied, are given in Table 16.5. These percentages are also applied for the opposing direction. The “breakpoint” in applicable segments corresponds to the northern section of the route through the city of Sisters and then the remainder of the route between Sisters and the city of Tumalo.

Additional Data Inputs for Intersections

The following is assumed about turning-movement volumes at these intersections:

• 10% left and right turns where there is an exclusive left-turn bay (Pine St and Locust St).
• 5% left and right turns where there is no exclusive left-turn bay (Larch St).
• The intersection with E Jefferson Ave does not include a leg on the northeast side of US-20. Thus, there are 0% left turns, and 5% right turns is assumed.

• 20% total turns (10% left, 10% right) at intersections with major roads.

Results

The segment LOS results are shown in Tables 16.6 and 16.7.

The facility LOS results are shown in Table 16.8.

Discussion

The overall facility LOS, based on a score of 1.98, is B. This is very close to the score threshold of 2.0 for LOS C. The facility average speed of 56 mi/h and FFS and threshold delays of 5% and <1%, respectively, indicate very good operations overall.

Individual segment LOS values are mostly B and C. The two-lane highway passing zone segments are mostly LOS B. Average speeds along the two-lane highway segments outside of the rural developed areas (Sisters and Tumalo) are high (58–62 mi/h). Both two-lane highway passing lane segments are LOS A.

Table 16.6. US-20 segment LOS analysis results, Part 1.

Volume-to-capacity ratios generally range from 0.2 to 0.35. The maximum d/c of 0.529 occurs in last two-lane highway (passing constrained) segment in Tumalo, as a result of the significant demand volume increase in this area. The roundabout intersection in Sisters has a d/c of 0.4 and a control delay of approximately 8 s/veh for the major street through movement, corresponding to an LOS of A (0–10 s/veh = LOS A).

The average speed along the arterial through Sisters is 30.6 mi/h. This is well above the posted speed limit—it is the correct value per the urban streets methodology, but this kind of result is a noted criticism of the HCM7 methodology. Since the intersections along the arterial segment through Sisters are all TWSC intersections, a d/c value is not reported. This is because the effects of the intersections are modeled as access points along the links (i.e., subsegments) and not modeled directly with the TWSC methodology, where delays due to left turns, for those intersections without separate left-turn bays, may be more accurately modeled.

Local observations of the operations, particularly at the TWSC intersections, should be made to determine the most appropriate way to analyze this stretch of roadway through Sisters.

According to the guidelines in Section 2.4, Hot-Spot Identification, no segments within the facility would be considered hot spots. For much higher demand volume scenarios, such as Labor Day weekend, several two-lane highway segments would likely reach LOS E or F. The arterial segment through Sisters may also become a hot spot under this demand volume condition, especially if left-turn movement volumes are moderate or higher. Under these conditions, it may be

Table 16.7. US-20 segment LOS analysis results, Part 2.

Table 16.8. US-20 facility analysis results.

necessary to restrict left turns from US-20 through Sisters, where no separate left-turn bay is provided. For a more permanent traffic growth situation, it may be necessary to consider converting the TWSC intersections to an alternate intersection form, such as a roundabout. See the Ohio US-42 Case Study for an example of testing different demand volume conditions on the route.

16.3 Reliability

It should be noted that in the US-20 (Oregon) sample data, the vendor’s sample size was not sufficient to produce 1-minute-resolution data. Although the data file reports for each minute of the day, the travel time column remains constant for long periods, indicating that data for those minutes is being imputed with historical data or modeled sources.

The reliability analysis focused on a 6-mile stretch of US-20 in the rural town of Sisters, Oregon (see Figure 16.14). For the analysis, the facility was divided into 10 segments with each segment measuring approximately 0.6 miles in length.

The facility was evaluated using five data analysis and visualization techniques that convert the raw speed and travel time data into charts and graphics for analysis and interpretation. These methods are described in Section 3.6).

Speed Heatmaps

Speed heatmaps for the Oregon US-20 case study are shown in Figure 16.15.

Figure 16.15 (eastbound speed from April 2021 through October 2021) shows generally high speeds (green) and consistent speeds (uniform color) outside of Sisters. The segments in downtown Sisters show slower speeds (colors moving to yellow and red), suggesting a lower speed limit, and particular slow speeds in the roundabout segments. These slowdowns can be categorized as recurring congestion, with every day showing similar slowdowns during the same hours of the day. The only clear evidence of nonrecurring congestion in Figure 16.15 is a few isolated red (slow) speeds, for example in the late May 2021 timeframe (potentially Memorial Day traffic).

Nonrecurring congestion is more evident in the days spanning from Christmas to New Year’s Day. This week shows notably lower speeds throughout the entire analysis corridor. The speed profile alone does not provide an obvious reason for why these speeds are slower during that particular month or two, but it is likely due to higher-than-usual demand for recreation during those days. Other causes of nonrecurring congestion include work zones, but it is unlikely to have construction in the mountains of Oregon in December and January. Nonrecurring congestion due to incidents or special events would be unlikely to last more than a day or two. As such, the general pattern on the speed heatmap allows us to infer that the US-20 corridor was likely experiencing a seasonal spike in demand that increased its travel times.

Speed Difference Heatmaps

Speed difference heatmaps for the Oregon US-20 case study are shown in Figure 16.16. The speed difference heatmap normalizes the display relative to the estimated FFS for the segment (using the 85th percentile speed) making it much easier to visually infer periods of recurring and nonrecurring congestion. In addition to the previously noted recurring congestion in the downtown area of Sisters and the nonrecurring congestion in the December/January timeframe, the speed difference heatmap shows periods of apparent unreliable travel during nighttime conditions. There is a potential bias here with low sample sizes on rural segments. For example, the low speeds could be due to a single (slow-moving) truck being the only source of probe data, which then translates into a slow recorded speed. However, low sample size is expected to be less of an issue when a segment shows congestion for multiple periods, which may then more likely be attributable to an incident, special event traffic, or severe weather event.

Box-and-Whisker Speed Plots

In the box-and-whisker weekly speed plots shown in Figure 16.17, speed (mi/h) is shown on the x-axis, and time (in weeks) is shown on the y-axis. The chart uses box-and-whisker plots to provide a summary of the weekly speeds for the analysis period.

Speed Confidence Band

The speed confidence plots for the Oregon US-20 case study are shown in Figure 16.18. The speed confidence bands highlight the reliability of speeds for different hours of the day. A “tight” speed band with 15th, 50th, and 85th percentile speeds all close together suggests a very reliable speed performance for that segment in the given period. A wider speed band with the 15th and 85th percentiles further away from the median suggests less reliable performance. A speed band that generally drops in speed (but remains tight) suggests that the facility is “reliably congested.” This is evident in downtown Sisters, which shows the speed band dropping as the downtown area becomes congested, but the travel is never unreliable (it is consistently slow).

All segments (except in downtown Sisters) display reliable speed confidence throughout the day during the April 2021 to October 2021 period between 11:00 AM to 5:00 PM. The speed band misses the date/calendar dimension but points to speed variability in the highlighted 11:00 AM to 6:00 PM timeframe. From here, the analyst can investigate the speed difference plot to see that congestion is most evident from roughly May through October (i.e., the high-tourism summer months), but less during the winter months.

TTI

The TTI plots for the Oregon US-20 case study are shown in Figure 16.19. The general shape of the cumulative TTI distribution allows inference on the variability of travel times along the segment. A generally steep TTI distribution suggests a generally reliable segment, while a more spread-out or flat distribution suggests a less reliable segment.

Summary

This case study presented a data-driven approach to explore travel time reliability on a rural highway in Oregon (i.e., US-20) as it traverses the small town of Sisters, popular recreational destinations, and rural areas. Five different visualization techniques were used to derive insights from higher-resolution vehicular probe data from INRIX XD.

This case study clearly shows how increased demand during holidays can affect the travel time reliability measures. In this case, popular recreational destinations saw a spike in demand during the Christmas and New Year’s Day holiday weeks, leading to higher travel times and degraded reliability over the analysis period.