ABSTRACT - University of Washington



FREIGHT DELAYS AT THE BORDER

The Impact of Homeland Security Alert Levels on Freight Delays at the Blaine, Washington International Truck Crossing

Abstract:

International borders often create trade barriers between countries. One of these barriers is due largely to the additional time and effort to transport goods through ports of entry. Recent developments such as the North American Free Trade Alliance (NAFTA) and World Trade Organization (WTO) efforts have helped reduce the effect of these barriers. As a result, the volume of international trade between Canada and the US grew substantially in the past decade and is currently the largest international trade flow in the world. In 2001, the terrorist attacks of September 11 (9-11) stimulated the US government to tighten security of goods and people across all of its borders, including Canada. These two factors, the expanded volume of trade along with tightened security, have led to increased delays for carriers moving goods across the border. This paper describes how much the delays have increased as a result of 9-11 and the influence that the Department of Homeland Security (DHS) Alert Level has on delay. The data presented is from the Pacific Highway Truck Crossing at Blaine, Washington on the US/Canadian Border. The average delay and variance of the delay for post 9-11 is examined and compared to pre 9-11 delay. It is shown that delay and variability has increased substantially and that adding additional processing booths is a reasonable course of action to reduce average delay and variance by reducing queue lengths.

David Kieninger, Master’s Degree Candidate

Department of Civil and Environmental Engineering

University of Washington

June 7, 2007

TABLE OF CONTENTS

Abstract: 1

Background 3

Probe Vehicle Data Set 6

Analysis of the Probe Vehicle Data Set 7

Alert Level Analysis 15

Compare to pre 9-11 19

Conclusions 20

Further Research 21

References 22

Acknowledgements 23

FIGURES

Figure 1: Northbound Travel Times 7/11/05 to 7/9/06 8

Figure 2: Southbound Travel Times 7/11/05 to 7/9/06 8

Figure 3: NB Hourly Breakdown 10

Figure 4: NB Breakdown by 15 Minute Period for Less than 1 Hour Delay 10

Figure 5: SB Hourly Breakdown 11

Figure 6: SB Breakdown by 15 Minute Period for Less than 1 Hour Delay 11

Figure 7: NB Delay by Month 12

Figure 8: SB Delay by Month 12

Figure 9: NB Delay by Day of the Week 13

Figure 10: SB Delay by Day of the Week 13

Figure 11: NB Delay by Hour 14

Figure 12: SB Delay by Hour 14

Figure 13: NB Orange vs. Yellow Alert 17

Figure 14: SB Orange vs. Yellow Alert 17

Figure 15: 95% Confidence Intervals for Alert Level Analysis Periods 18

Figure 16: 95% Confidence Intervals for Alert Level Analysis Periods 18

Figure 17: Average Delay Before and After 9/11 20

TABLES

Table 1: Summary of NB and SB Delay 20

Background

Trucking volumes continue to increase across the Pac Highway Border Crossing at Blaine, WA. Along with this increased volume comes increased congestion. Increased border security measures since 9/11 have also contributed to congestion and delays for freight movers. The congestion has caused travel time reliability to suffer as well. Reliability is often cited by trucking companies as one of the most important measures for them of the transportation system. The purpose of this research is to examine the effects the Department of Homeland Security Alert Levels have on expected delay at the border as well as the variability of that delay. It is hoped that this information can be used by trucking companies to help them more effectively schedule their logistics depending on alert level. For example, they could know how many more trucks they would need to maintain a 2 hour delivery time if the alert level was raised from yellow to orange.

The original intent of this research was to see what the correlation is between Homeland Security Alert levels and average delays seen by trucks crossing the US/Canadian Border. Canada remains the US’s largest trading partner. The truck crossing at Blaine, WA is the 4th busiest commercial crossing on the US/Canadian border. It was expected that average delays would increase as the alert level increased due mainly to tighter security and more thorough inspections along with the resulting queues that would develop as a result of longer processing times.

Picture 1 is a map of the roads in the vicinity of the Pac Highway Truck Crossing taken from Mapquest [7]. Of note is that trucks traveling NB on I-5 are required to exit the freeway about 1 mile before the border at exit 275 shown on the map and use SR 543 to get to the commercial crossing. The Peace Arch Crossing on I-5 is for passenger vehicles only. There is a similar situation for trucks driving south from Canada.

[pic]

Picture 1: Canadian Border in the Blaine Vicinity [7]

Picture 2 shows the WSDOT and BC traveler information websites for these crossings [5,6]. The detection areas shown in green give a good idea of the types of queues that frequently occur on the approaches to these crossings. For example, when queues spill back to I-5 exit 275, the green will show as black on the WSDOT website.

[pic][pic]

Picture 2: WSDOT and BCMOT Traveler Information Websites [5,6]

Picture 3 is an aerial photo facing north [2]. It shows more of the details of the border crossing layout. The trucks at the top of the picture are queued to go over to the US Customs and Border Patrol (CBP) inspection area just to the left of the picture. There is a circular area in the center of their facility for handling buses. Next to that are the passenger vehicle booths. The NB Canadian inspection facilities follow a similar layout. The SB facility has 2 general purpose lanes and 1 FAST lane. The NB facility has 2 general purpose lanes.

[pic]

Picture 3: Overhead Photo Facing North [2]

Probe Vehicle Data Set

Part of the impetus for doing this research was the availability of a fairly large data set that had been provided by a private trucking company to the Washington State Transportation Research Center (TRAC) and the University of Washington. The trucking company makes daily deliveries of jet fuel between the US and Canada using the Pac Highway Truck Crossing. They have been collecting data on the delays experienced by their trucks for the past few years. Their trucks are equipped with GPS transponders that record their individual wait times at the border for each trip. The drivers hit a button to start a timer when they enter the queue to begin tracking their border crossing time and hit the button again when they have cleared the border. They have made that data available since July 2005.

This data set is referred to as the probe vehicle data set. The data set used is approximately 1 year worth of data from July 11, 2005 to July 9, 2006 and has about 11,250 trips recorded. The trucks are full when traveling Northbound (NB) and empty when returning Southbound (SB.) The company, trucks, and drivers are registered in the FAST system and are able to use the FAST lane when returning SB. Therefore, the SB data isn’t representative of all trucks going SB. It is only representative of trucks using the FAST lane.

It was expected that these wait times over an extended period could be compared to historic data for the alert levels. The historic alert level data could be easily obtained through the Department of Homeland Security’s website.

Analysis of the Probe Vehicle Data Set

The first step in analyzing the data set was to check it for errors. There was a wide variety of delays recorded as can be expected from the nature of the inspection processes used at the border. All vehicles are screened, but random sampling and flagging some vehicles for more intensive inspections would cause some delays to look like outliers in the data set. These outliers are an important part of the data set and shouldn’t be excluded from the analysis.

Still, it was apparent that there were some errors in the data. For instance, if a truck was shown as being delayed from 9 am to 2 pm going NB, but also had another record showing that it was delayed from 10 am to 11 am going SB on the same day, then there is obviously a problem with that data. To start with, the 5 data points larger than 4 hours of delay were checked and they all appeared to be in error using a test checking for the problem shown in the example above. From there, the next 80 of the highest delayed vehicles were manually checked by hand and less than 10 of them had apparent errors. Because of the significance of the 70 not found in error and because the entire data set was not rigorously error checked, it was decided that these points would remain in the analysis. The 5 bad points over 4 hours were removed along with 15 points that had delay of 0 seconds. This weakness in the analysis will be discussed in the recommendations for further research at the end of the paper.

Figures 1 through 12 attempt to summarize the probe data set and give a good background for the analysis that was done. NB and SB traffic is separated for all the analysis since there are different agencies controlling the delay. Also, it should be noted that all of the probe data for SB vehicles is for FAST approved trucks and drivers.

Figures 1 and 2 are plots of all the probe vehicle data from 7/11/05 to 7/9/05. There are 5686 data points for NB and 5544 data points for SB. They show the range and variation of delay experienced at the border. They also show the gaps where no data was available, from 12/5/05 to 1/31/06 and from 6/1/06 to 6/18/06.

[pic]

Figure 1: Northbound Travel Times 7/11/05 to 7/9/06

[pic]

Figure 2: Southbound Travel Times 7/11/05 to 7/9/06

Figure 3 shows the distribution of delay experienced for NB trucks by 1 hour increments. 96% of trucks experience less than 1 hour of delay. Figure 4 is a further breakdown of the less than 1 hour delay into 15 minute segments. 50% of those are less than 15 minutes of delay.

Figures 5 and 6 are the same breakdown, but for SB vehicles. The breakdown percentages are very similar.

Figures 7 and 8 show the distribution of delay by month for NB and then for SB trucks. The delay is characterized by the average delay and the standard deviation of delay. There are actually 2 bars for July since there was data from July of 2005 and July 2006.

Figures 9 and 10 show the distribution of delay by day of the week for NB and SB trucks. Again, the average delay and standard deviation of delay is plotted. It is interesting to see that for NB trucks, average delay drops by nearly half on weekends. But for SB trucks the average delay remains relatively constant all week.

Figures 11 and 12 show the distribution of delay by hour of the day for NB and SB trucks. The average delay and standard deviation of delay is plotted. Also, the number of data points per hour is shown because the late night hours have large variation due to the low samples in those hours. Another thing to note on this chart is that for SB traffic, the FAST lane becomes the general purpose lane from 11 pm to 6 am and is the only lane open.

[pic]

Figure 3: NB Hourly Breakdown

[pic]

Figure 4: NB Breakdown by 15 Minute Period for Less than 1 Hour Delay

[pic]

Figure 5: SB Hourly Breakdown

[pic]

Figure 6: SB Breakdown by 15 Minute Period for Less than 1 Hour Delay

[pic]

Figure 7: NB Delay by Month

[pic]

Figure 8: SB Delay by Month

[pic]

Figure 9: NB Delay by Day of the Week

[pic]

Figure 10: SB Delay by Day of the Week

[pic]

Figure 11: NB Delay by Hour

[pic]

Figure 12: SB Delay by Hour

Alert Level Analysis

Early in the research, it was discovered that the DHS alert level has not changed as often as originally assumed. The changes become even less significant when compared to the timeframe the probe vehicle data is available for. On a nationwide level, the alert level has remained at yellow for most of the 5 years since its inception. It has been elevated to orange 5 times during the last 5 years. Blue, green, and red have not been used on a nationwide level that would affect trucking. The other problem is that none of these changes has occurred during the timeframe that the data collected from the trucking company is available for. The trucking data is only available from 7/11/05 until 7/9/06. The following is a list of dates when the DHS alert level was changed.

• March 12, 2002 – Introduction of Homeland Security Advisory System At Yellow

• September 10, 2002 – Raised from Yellow to Orange

• September 24, 2002 – Lowered from Orange to Yellow

• February 7, 2003 – Raised from Yellow to Orange

• February 27, 2003 – Lowered from Orange to Yellow

• March 17, 2003 – Raised from Yellow to Orange

• April 16, 2003 – Lowered from Orange to Yellow

• May 20, 2003 – Raised from Orange to Yellow

• May 30, 2003 – Lowered from Orange to Yellow

• December 21, 2003 – Raised from Yellow to Orange

• January 9, 2004 – Lowered from Orange to Yellow

• August 1, 2004 – Raised from Yellow to Orange, specifically for the financial services sectors in New York City, Northern New Jersey, and Washington, DC

• November 10, 2004 – Lowered from Orange to Yellow, for the financial services sectors in New York City, Northern New Jersey, and Washington, DC

• July 7, 2005 – Raised from Yellow to Orange for mass transit

• August 12, 2005 – Lowered from Orange to Yellow for mass transit

• August 10, 2006 – Raised from Yellow to Red for flights originating in the United Kingdom bound for the United States; raised to Orange for all commercial aviation operating in or destined for the United States.

• August 13, 2006 – Lowered from Red to Orange for flights originating in the United Kingdom bound for the United States; remains at Orange for all domestic and international flights.

The one time that the alert level was changed once during this timeframe was specific to mass transit systems. The alert level was increased from yellow to orange on July 7, 2005 for mass transit, which is just before the probe vehicle data set begins. The alert level was dropped from orange back down to yellow on August 12, 2005, which is during the timeframe of the probe vehicle data set. Although the higher alert level was specific to mass transit and, in theory, would not affect trucks crossing the border, the data was available both during and after so it was compared. The data is summarized in Table 1.

It was interesting to see that for SB vehicles, the average delay during the orange period between 7/11/05 to 8/11/05 was about 28 minutes. This is 5 minutes longer than the 23 minute average delay found during the 4 yellow periods that it was compared to. These periods are shown in Figures 13 and 14. Figures 15 and 16 show the statistical significance of the differences found in delay. The NB change in delay doesn’t appear to be statistically significant, but the SB change is. Using a single change of the alert level, especially an alert that didn’t specifically or broadly address trucks crossing the border, makes drawing any hard conclusions difficult. Seasonal variations and other factors may be involved in this difference. But it does suggest that there is some influence and further research is warranted to determine what the correlation is between alert levels and average delays.

[pic]Figure 13: NB Orange vs. Yellow Alert

[pic]

Figure 14: SB Orange vs. Yellow Alert

[pic]

Figure 15: 95% Confidence Intervals for Alert Level Analysis Periods

[pic]Figure 16: 95% Confidence Intervals for Alert Level Analysis Periods

Compare to pre 9-11

Since the alert level variations didn’t match up well with the existing probe vehicle data set, the research was then directed toward comparing post 9/11 delays to pre 9/11 delays. FHWA commissioned a study to benchmark border crossing delays in 2001 [1,2]. The data from Blaine, WA was collected during a 3 day period prior to 9/11, specifically July 10-12, 2001. That study found average crossing time for northbound (NB) vehicles was 21.0 minutes and for southbound (SB) was 17.3 minutes. The 95% crossing times were 35.3 minutes NB and 35.6 minutes SB. These benchmarked travel times will be used for the comparison. It should be noted that for the purposes of this paper, delay is defined as the total travel time spent crossing the border. This total travel time is the effective delay experienced by the carrier as a result of crossing the international boundary. The FHWA study had different objectives related to delay so they calculated delay by subtracting the free-flow travel time from the average travel time and is not representative of the total crossing time.

Because the probe vehicle data set only included trucks using the FAST lane, there was no data available for the non-Fast general commercial vehicles. To fill this gap, data from a study conducted by Whatcom County Council of Governments (WCOG) in 2006 was used to get a representative sample of the delay experienced by SB non-FAST trucks [3,4]. That study used data from the probe vehicle data to validate that their data was representative of FAST vehicles and could be reasonably expected to be representative of non-FAST vehicle delays.

Figure 17 and Table 1 summarize the comparison between pre 9/11 delay and post 9/11 delay. The average delays found for post 9/11 delay using the probe vehicle data set is 22 minutes NB and 24 minutes SB. The standard deviation of the delay is 20 minutes NB and 24 minutes SB. The WCOG data indicates that travel times for general SB traffic has increased substantially since 9/11 to 1 hour and 23 minutes. The SB data is of interest because it represents the largest difference between pre 9/11 and post 9/11. This was expected because the US reaction to the terrorist attacks was to dramatically increase security and inspections. Also, the SB FAST lane data after 9/11 compares favorably to the benchmark data for pre 9/11 delay. This indicates that the FAST lane program has been successful in reducing delay. There is still much that could be done to address the significant delays for the general commercial vehicles that aren’t able to use the FAST lane.

[pic]

Figure 17: Average Delay Before and After 9/11

[pic]Table 1: Summary of NB and SB Delay

Conclusions

The delays experienced at the border do not seem to be significant for NB or for SB FAST trucks. The average delays are near 20-25 minutes in each direction and this would seem to be an acceptable level of delay for a border crossing. This delay should not be considered a significant impact to the operating costs of most freight carriers or to the economy of either country. The variability of the delay, with standard deviations of approximately 20-25 minutes, could be considered a mild impact to the operating cost of individual carriers with otherwise short travel times. This variability is not significant for longer haul trucks when compared to the multitude of other variability in their trips. Also, the long delays of over 1 or 2 hours, while frustrating, appear to be rare enough to not be a significant impact on costs.

The delays experienced by non-Fast trucks of nearly 1.5 hours is a significant issue and is a negative impact on the operating costs of carriers and the economy of both countries. It is much longer than the benchmark delays from 2001 and does not seem to be an acceptable delay. It may be a barrier to trade between these two countries.

Department of Homeland Security Alert Levels have not changed significantly since the program was introduced. There was only one alert level change during the 1 year analysis period. The alert level change was specific to mass transit systems but it had a statistically significant impact on SB truck delays at the border. The implementation of higher security corresponding to the terrorist attacks of September 11, 2001 was also evaluated and shows significant increases in delay for SB non-FAST trucks. The data shows that average delay for SB freight to the US has increased compared to pre 9-11 conditions.

US Customs and Border Patrol (CBP) is the branch of the Department of Homeland Security that acts as the first line of defense for our nation’s borders. Their first priority is to defend and protect the borders of this country. Any and all improvements that can be done to reduce delay are subject to meeting this priority first. Variability in processing time is inherent to the random sampling and inspection processes that CBP performs. US Customs has made many advances in their processing and inspection procedures such as FAST, CTPAT, ACE, etc. which has helped lower the average delay and variability costs for carriers.

A secondary source of delay and variability is the long queues that often form at the border. These queues are a result of capacity limitations. With tighter security measures and more rigorous inspections, the effective capacity of each inspection booth has been reduced. Little if any infrastructure, such as booths and lanes, has been added since 9/11 to make up for the lost capacity. More booths would allow greater flexibility for US Customs officials to respond to peak demands and reduce queues. The additional throughput would lead to lower average delays and lower variability by addressing the secondary source of delay, excessive queue lengths.

Further Research

The probe vehicle data continues to be sent to TRAC and the University of Washington. This will continue to create a larger data set that should eventually be large enough to catch the variations in DHS alert levels.

More thorough, automated error checking and data cleaning could be completed on the existing probe vehicle data set to make it more reliable and useful.

Currently, there is a two-tiered lane assignment for the SB booths, FAST and non-Fast. Continued advances in technology could be evaluated to make better use of the CTPAT and ACE systems. These systems could be used to pre-screen trucks into three or more tiers of inspection levels. This would reduce delay for all the lower inspection tiers because they wouldn’t be in queue as long.

The feasibility of adding more lanes, especially SB, should be evaluated. The tightened security measures since 9/11 have effectively reduced the service rate and capacity of each general inspection booth. Short of changing inspection procedures, more booths will need to be added in order to reduce the SB delays back to the benchmark delays of 2001.

References

1. U.S. Department of Transportation (USDOT), Federal Highway Administration (FHWA) 2002a. Evaluation of travel time methods to support mobility performance monitoring: International border crossing truck travel time for 2001. Texas Transportation Institute and Battelle Memorial Institute.

2. U.S. Department of Transportation (USDOT), Federal Highway Administration (FHWA) 2002b. Evaluation of travel time methods to support mobility performance monitoring: Border crossing freight delay data collection and analysis 2001 data collection-Pacific Highway (Blaine Border) crossing. Battelle Memorial Institute.

3. Goodchild, Anne, Steven Globerman, Susan Albrecht. 2006. Service time variability at the Blaine, Washington international border crossing and the impact on regional supply chains. Working Paper.

4. Halcrow Consulting, 2006. Whatcom Council of Governments Pacific Crossing Commercial Vehicle Border Delay Survey Final Report.

5. WSDOT Canadian Border Traffic



6. Government of British Columbia - Advanced Traveler Information System



7. Mapquest



Acknowledgements

I would like to thank the following people and institutions for their assistance in this research.

Anne Goodchild, PhD, Faculty Advisor

Department of Civil and Environmental Engineering

University of Washington

Edward McCormack, PhD

TRAC

Hugh Conroy

Whatcom Council of Governments

WSDOT

TRANSNOW

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download