Measuring Border Delay and Crossing Times at the US–Mexico Border
Final Report
Automated Crossing and Wait Time Measurement
CHAPTER 8: EVALUATION OF CROSSING AND WAIT TIMES DATA
This chapter highlights key findings from tests performed to evaluate the accuracy of wait time and crossing time data measured by the system. Evaluation tests were performed at different times using techniques such as manual license plate reading and GPS probes. The following evaluation tests were performed, and results from the tests are described in subsequent sections:
Evaluation tests performed at BOTA in August 2009 by comparing RFID data with manual license plate data collection at BOTA immediately following installation of crossing time measurement system.
Evaluation tests performed at BOTA in January 2012 by comparing RFID-measured wait times and crossing times with data measured by GPS installed inside US-bound trucks heading into the United States.
Evaluation tests performed at Pharr-Reynosa in May 2011 by comparing RFID-measured wait times with data measured by GPS installed inside US-bound trucks heading into the United States.
Evaluation Performed at BOTA in August 2009
In August 2009, the first round of evaluation tests was performed immediately after installation of the initial two RFID reader stations at BOTA. The objective of the evaluation was to determine accuracy of crossing times determined by the RFID system, by comparing them with crossing time measured in the field using video data collection of truck license plates.
Portable and digital video cameras were installed at several locations along the truck path at BOTA. Using video recordings and manual observation, the presence of tags were noted and license plates of the trucks that had the tags were read. Based on these two observations, the volume of trucks and number of tags read at RFID stations were analyzed. Two handheld cameras were installed at pre-defined locations close to RFID reader sites.
The following tests were conducted to measure the accuracy of crossing time of US-bound trucks at BOTA:
- Compare the total number of tags read by RFID system to total tags observed in the field.
- Identify the cause of disparity between number of tags read by RFID stations on the Mexico and US sides of the border.
- Compare average truck crossing times observed in the field and determined by the RFID system.
Using video recording of trucks going under both RFID stations on the US side, tags affixed to the windshield were identified and counted and that number compared with the total number of tags read by the RFID. Trucks carry tags issued by several agencies and carry stickers related to the permitting process. Since some of these stickers have a similar appearance to RFID tags, there are instances of overestimating the number of RFID tags carried by trucks. Thus, even though there is no guarantee that the tags observed in the field can be called “readable tags” or just stickers, field evaluation still indicated the capability of the RFID system to read an adequate number of tags. From video recordings, the number of tags attached to windshield of trucks going under the RFID stations were counted and the total tags read were compared with the total read by the RFID system, which were obtained from the central database. A detailed (15-minute) breakdown of the number of tags visually counted and read by the RFID reader was determined for the Mexico and the US side. An average 81percent of total tags were read by the RFID readers on the US side and 95 percent of the total tags were read by the reader on the Mexico side.
In addition, the total number of tags read by readers against the total number of trucks passing both RFID stations was obtained. Table 10 shows the penetration rates or percentage of total US-bound trucks successfully identify by the readers at both stations. The results show that the penetration rate is above 90 percent on the Mexico side and above 75 percent on the US side.
| Date | Time Period | Location | Total Number Trucks Counted at RFID Station on the MX Side | Total Number Tags Read By the RFID Reader | Penetration Rate (in percent) |
|---|---|---|---|---|---|
| 08/19/09 | 7:00-9:00AM | MX | 230 | 206 | 90 |
| 08/19/09 | 3:00-5:00PM | MX | 161 | 148 | 92 |
| 08/20/09 | 7:00-9:00AM | MX | 231 | 218 | 94 |
| 08/18/09 | 4:30-5:30PM | US | 153 | 109 | 71 |
| 08/19/09 | 7:30-9:30AM | US | 224 | 180 | 80 |
| 08/19/09 | 4:30-5:30PM | US | 229 | 166 | 72 |
| 08/20/09 | 7:30-9:30AM | US | 217 | 161 | 74 |
Analysis of archived RFID data at BOTA showed that a significant number of tags read by the RFID station on the Mexico side are not read by the station installed at the exit of the DPS facility on the US side. The total number of tags read by the RFID reader on the Mexico side is on an average 30 percent higher than the total number of tags read by the reader on the US side. This disparity is illustrated in figure 59.
This discrepancy was considered most likely because there are several driveways and at least one major roadway between the RFID station and the entrance to the Mexican Customs. Thus, trucks go under the Mexican RFID station, but instead of going to the Mexican Customs compound they go into side roadways and not under the reader on the US side. The researchers believe that this discrepancy does not hamper collection of sample size and estimation of average crossing times. In fact, there is an adequate sample size to estimate the average crossing times of trucks reliably.
Finally, using video recordings of trucks over several hours at RFID stations and matching the license plate numbers of trucks manually collected at both stations, average crossing times of trucks observed in the field were compared to those calculated by the RFID system. The average truck crossing times were calculated at 15-minute intervals using the last two hours of data (referenced from the time when the tag was first read by the reader on the US side). Table 11 shows a comparison of average crossing time determined by license plate readings and by the RFID system. The results show that the average crossing time relayed by the RFID system is quite comparable to the ones computed in the field for most of the periods.
On average, the difference in crossing time estimated by the RFID system and the manual license plate reading method was 4.7 minutes for all time periods listed in table 11. The original minimum value was set at 3 minutes, which was based on a method of comparing crossing times of trucks carrying both GPS devices and RFID tags. In this case the trucks, whose license plates were identified on both sides of the border, may or may not have RFID tags. Hence, while calculating the average crossing time using license plate reads and the RFID system, there is no guarantee that the same trucks are being used in the calculation. Due to this reason, average crossing times estimated by the RFID system may differ with that estimated using manual license plate reads.
| Date | Time of Calculation on the US Side | Time Window Used to Calculate Average Crossing Time | Average Crossing Time from the RFID System | Average Crossing Time from the Field Data Collection |
|---|---|---|---|---|
| 08/19/09 | 9:00AM | 8:00 – 9:00 | 33.67 | 32.79 |
| 9:15AM | 8:15 – 9:15 | 31.35 | 34.50 | |
| 9:30AM | 8:30 – 9:30 | 34.90 | 44.33 | |
| 5:00PM | 4:00 – 5:00 | 43.00 | 48.07 | |
| 5:15PM | 4:15 – 5:15 | 41.08 | 48.02 | |
| 5:30PM | 4:30 – 5:30 | 37.02 | 49.07 | |
| 08/20/09 | 9:00AM | 8:00 – 9:00 | 49.28 | 47.10 |
| 9:15AM | 8:15 – 9:15 | 47.33 | 45.79 | |
| 9:30AM | 8:30 – 9:30 | 46.83 | 45.71 |
Evaluation Performed at BOTA in January 2012
In January 2012, an evaluation of wait time and crossing time from the BOTA system was performed using three GPS devices installed on trucks operated by STIL Inc., a drayage carrier based in Ciudad Juarez. The GPS units provide longitude, latitude, and timestamp at 30-second intervals and stored the data in the unit. The carrier was not asked to run its trucks on a particular schedule. The GPS units were attached to the carrier’s trucks for two weeks. However, wait time and crossing time data collected were not continuous, since trucks were running on their own schedule.
The data obtained from GPS units were plotted in GIS software and overlaid over local roadway network and RFID zones, which are polygons surrounding RFID stations as illustrated in figure 60. Wait times and crossing times of trucks were measured by taking the time differences between when trucks exited starting zones and entered ending zones. Median value of wait times and crossing times of trucks measured by RFID was obtained approximately at the same time trucks exited the zone surrounding the first RFID reader station in Mexico. Table 12 and table 13 show the comparison of GPS- and RFID-measured wait and crossing times.
The absolute average difference for wait time was 5.5 minutes (11 percent) and for crossing time was 7.6 minutes (15 percent). This minor discrepancy was because GPS points do not always appear exactly at the location of the RFID station location, and also because it was not known whether the trucks carrying GPS devices also had RFID transponders that could have been read by the RFID system.
| Trip Date | GPS Unit | Time Truck Crossed R1 | Time Truck Crossed R2 | GPS Wait Time (in minutes) | Median RFID Wait Time (in minutes) | Absolute Difference (in minutes) | Difference (in percent) |
|---|---|---|---|---|---|---|---|
| 1/16/2012 | 10401 | 9:16:12 | 9:27:43 | 12 | 12 | 0.0 | 0 |
| 1/16/2012 | 10401 | 16:17:12 | 16:43:15 | 26 | 28 | 2.0 | 8 |
| 1/17/2012 | 10401 | 10:16:29 | 11:25:38 | 71 | 73.5 | 2.5 | 4 |
| 1/17/2012 | 10401 | 15:38:34 | 16:11:08 | 33 | 27.5 | 5.5 | 17 |
| 1/19/2012 | 10401 | 12:04:40 | 12:29:20 | 25 | 23 | 2.0 | 8 |
| 1/21/2012 | 10401 | 10:31:00 | 11:16:00 | 45 | 36.5 | 8.5 | 19 |
| 1/23/2012 | 10401 | 11:21:19 | 12:02:00 | 25 | 21.5 | 3.5 | 14 |
| 1/24/2012 | 10401 | 11:26:00 | 13:06:00 | 102 | 127 | 25.0 | 25 |
| 1/26/2012 | 10401 | 9:30:40 | 10:51:00 | 80 | 88 | 8.0 | 10 |
| 1/16/2012 | 30302 | 7:43:30 | 8:15:24 | 32 | 27 | 5.0 | 16 |
| 1/17/2012 | 30302 | 14:05:35 | 14:59:15 | 54 | 50.5 | 3.5 | 6 |
| 1/16/2012 | 30369 | 7:57:45 | 8:21:10 | 24 | 23 | 1.0 | 4 |
| Trip Date | GPS Unit | Time Truck Crossed R1 | Time Truck Crossed R3 | GPS Crossing Time (in minutes) | Median RFID Crossing Time (in minutes) | Absolute Difference (in minutes) | Difference (in percent) |
|---|---|---|---|---|---|---|---|
| 1/16/2012 | 10401 | 9:16:12 | 9:42:33 | 26 | 18 | 8.0 | 31 |
| 1/16/2012 | 10401 | 16:17:12 | 16:45:25 | 28 | 31 | 3.0 | 11 |
| 1/17/2012 | 10401 | 10:16:29 | 11:24:43 | 68 | 72.5 | 4.5 | 7 |
| 1/18/2012 | 10401 | 9:55:35 | 11:01:46 | 66 | 74 | 8.0 | 12 |
| 1/18/2012 | 10401 | 15:59:58 | 16:31:00 | 31 | 40 | 9.0 | 29 |
| 1/19/2012 | 10401 | 12:04:40 | 12:30:13 | 26 | 29.5 | 3.5 | 13 |
| 1/21/2012 | 10401 | 10:31:00 | 11:49:00 | 78 | 65 | 13.0 | 17 |
| 1/23/2012 | 10401 | 11:21:19 | 12:10:51 | 34 | 27 | 7.0 | 21 |
| 1/24/2012 | 10401 | 11:26:00 | 13:11:25 | 105 | 106 | 1.0 | 1 |
| 1/26/2012 | 10401 | 9:30:40 | 11:50:00 | 139 | 118 | 21.0 | 15 |
| 1/16/2012 | 30302 | 7:43:30 | 8:23:30 | 40 | 33 | 7.0 | 18 |
| 1/17/2012 | 30302 | 14:05:35 | 15:20:00 | 74 | 81 | 7.0 | 9 |
| 1/16/2012 | 30369 | 7:57:45 | 8:36:00 | 48 | 41 | 7.0 | 15 |
Evaluation Performed at PHARR-REYNOSA in May 2011
In May 2011, an evaluation of wait time of trucks from the Pharr-Reynosa system was performed using three GPS devices installed on trucks operated by a drayage carrier based in the region. The carrier was not provided the GPS units; rather they were installed by the carrier for its internal purposes. The carrier provided archived GPS log data which include longitude, latitude, and timestamps at sporadic intervals. The carrier was not asked to run trucks on a particular schedule. Hence, wait time data collected were not continuous since trucks were running on their own schedule.
The data obtained from the carrier were plotted in a GIS software and overlaid over local roadway network and RFID zones, which are polygons surrounding RFID stations. Wait times of trucks were measured by taking the time difference between when trucks exited starting zones and entered ending zones. Median value of wait times of trucks measured by RFID was obtained approximately at the same time trucks exited the zone surrounding the first RFID reader station in Mexico. Table 14 shows the comparison of GPS and RFID measured wait times of US-bound trucks.
The absolute average difference for wait time was 3.6 minutes (8 percent). This minor discrepancy is because GPS points do not always appear exactly at the location of the RFID station location, and also because it was not known whether the trucks carrying GPS devices also had RFID transponders that could have been read by the RFID system.
| Trip Date | Trip No. | Time Truck Crossed R1 | Time Truck Crossed R3* | GPS Wait Time (in minutes) | Median RFID Wait Time (in minutes) | Absolute Difference (in minutes) | Difference (in percent) |
|---|---|---|---|---|---|---|---|
| 17-May-11 | 1 | 12:01:01 | 13:16:49 | 75.9 | 76 | 0.10 | 0 |
| 17-May-11 | 2 | 14:48:00 | 15:38:01 | 50 | 53 | 3.00 | 6 |
| 17-May-11 | 3 | 16:20:28 | 16:37:24 | 17 | 19 | 2.00 | 12 |
| 17-May-11 | 4 | 17:54:07 | 18:19:27 | 25.3 | 27 | 1.70 | 7 |
| 17-May-11 | 5 | 19:01:22 | 19:21:45 | 20.4 | 18 | 2.40 | 12 |
| 17-May-11 | 6 | 20:30:02 | 20:50:23 | 20.4 | 18 | 2.40 | 12 |
| 17-May-11 | 7 | 20:37:59 | 20:53:11 | 15.2 | 15 | 0.20 | 1 |
| 18-May-11 | 1 | 13:56:50 | 16:16:51 | 140 | 123 | 17.00 | 12 |
| 18-May-11 | 2 | 14:12:46 | 16:09:16 | 116.6 | 117 | 0.40 | 0 |
| 18-May-11 | 3 | 14:14:55 | 16:18:16 | 123.4 | 121 | 2.40 | 2 |
| 18-May-11 | 5 | 14:55:27 | 17:02:12 | 126.7 | 139 | 12.30 | 10 |
| 18-May-11 | 6 | 16:02:45 | 17:03:34 | 60.9 | 64 | 3.10 | 5 |
| 18-May-11 | 7 | 16:47:22 | 17:27:21 | 40 | 40 | 0.00 | 0 |
| 18-May-11 | 8 | 17:22:44 | 18:03:39 | 40.9 | 43 | 2.10 | 5 |
| 18-May-11 | 9 | 19:01:33 | 19:19:09 | 17.6 | 16 | 1.60 | 9 |
| 18-May-11 | 10 | 19:43:04 | 20:08:28 | 25.5 | 30 | 4.50 | 18 |
| 18-May-11 | 11 | 19:47:37 | 20:09:15 | 21.7 | 19 | 2.70 | 12 |
| 20-May-11 | 4 | 14:29:40 | 14:59:27 | 29.8 | 34 | 4.20 | 14 |
| 20-May-11 | 5 | 15:06:54 | 15:42:31 | 35.6 | 37 | 1.40 | 4 |
| 20-May-11 | 6 | 16:47:37 | 17:13:00 | 25.4 | 31 | 5.60 | 22 |
| 20-May-11 | 7 | 16:47:37 | 17:14:28 | 26.9 | 31 | 4.10 | 15 |
| 20-May-11 | 8 | 17:00:19 | 17:26:01 | 25.7 | 30 | 4.30 | 17 |
| 21-May-11 | 1 | 9:21:08 | 10:36:59 | 75.9 | 81 | 5.10 | 7 |
| 21-May-11 | 4 | 11:54:48 | 13:15:59 | 81.2 | 89 | 7.80 | 10 |
| 21-May-11 | 5 | 12:17:15 | 14:02:16 | 105 | 106 | 1.00 | 1 |
* = Pharr-Reynosa international border crossing has one additional RFID reader station prior to CBP primary. Hence, the reader station at CBP is numbered R3 instead of R2. ↑
previous | next
