Identifying schemes to dodge taxes or steal fuel—and finding solutions—will bolster revenues for constructing, maintaining, and operating highways and bridges.

Excise taxes on motor fuels account for the majority of the funding for Federal and State transportation programs. Fuel taxes collected are used to construct and maintain roads and bridges in addition to helping fund mass transit projects. In 2014, these taxes amounted to more than $80 billion. That revenue could be even greater, if taxation agencies can find a way to put an end to fuel tax evasion.

Collection of motor fuel taxes in the United States occurs at various transaction levels. Federal fuel taxes are assessed when the fuel is removed from a bulk terminal, generally when it is loaded onto delivery trucks. A little more than half of the States employ that same point of taxation. The remainder levy the tax as a part of subsequent transactions through to the final purchaser.

The Federal Government and most States generally compute the rate of taxation on a cents-per-gallon basis. But some States impose the taxes as a percentage of the cost. The Internal Revenue Service (IRS) and the State taxation agencies employ various processes to collect these taxes.

The most familiar forms of transport for motor fuels are tank trailers, which usually pick up loads of fuel at bulk terminals and transport them to retail service stations or other storage locations. These tank vehicles can transport between 7,500 and 8,500 gallons of fuel, and the Federal and State taxes on such product volumes can easily exceed $5,000. Because the volume of motor fuel sales in the United States exceeds 250 billion gallons each year, even a small amount of tax evasion can add up to significant lost revenue, possibly as much as $1 billion.

Over the years, the States and the IRS have worked together to try to eliminate fuel tax evasion. They have modified collection processes to reduce the opportunities for evasion, but some schemes remain difficult to detect. Because many are probably missed, the level of evasion is not easily measured.

The Federal Highway Administration has coordinated the efforts of the IRS and States and provided some funding to combat this problem. Even so, the taxation agencies cannot afford to employ initiatives to combat evasion unless they are confident that the processes will be successful.

In 2009 the National Transporta-tion Research Center of the Oak Ridge National Laboratory in Tennessee submitted a proposal to FHWA’s Exploratory Advanced Research Program to consolidate innovative software with existing technologies into a tool that could provide information on a tanker’s travel and operations. The Oak Ridge researchers titled the initial work Supply Chain-Based Solution to Prevent Fuel Tax Evasion.

“The development of a tracking system capable of capturing transactional data from the time a load of fuel leaves the terminal until it is delivered to the customer [would help] auditors identify and research questionable activity and greatly reduce the probability of fuel tax evasion at the terminal level,” says Dawn Lietz, deputy administrator of the Motor Carrier Division of the Nevada Department of Motor Vehicles.

Addressing Evasion Schemes

Starting in the mid-1980s, the IRS and State revenue agencies identified several fuel tax evasion schemes and tackled them through measures such as moving the point of taxation to an earlier time in the distribution chain for gasoline (1986) and diesel (1994), and by adding a red dye (1994) to the diesel fuel intended for use only off the highway. However, a number of ways remain for the tax to be evaded.

One form of evasion, which can be difficult to identify, is the adulteration of previously tax-paid product. In this case, a driver might stop along a delivery route, unload some of the good fuel, and replace it with nonmotor fuel, such as used motor oil or other waste liquids. In small concentrations, this adulteration is not easy to detect. One aspect of Oak Ridge’s proposal was to use various technologies to prevent adulteration by nontaxed liquids.

Another piece of the Oak Ridge proposal was to track the movement of tankers to ensure that operators report the proper delivery location. States have different tax rates, all levied on a cents-per-gallon basis. These differences can be as much as 56 cents per gallon. If a tanker operator picks up a load of fuel in a State with a lower tax rate and reports it as delivered in that State, but then delivers it to a location in another State with a higher tax rate, and does not report that diversion, the offender can profit from the difference in tax rates.

makes the loading process relatively easy.

To address these schemes, the major feature of Oak Ridge’s proposed system was the creation of software that would employ evidential reasoning analysis to assess whether vehicles are being operated according to route expectations, with allowances for certain exceptions. Evidential reasoning analysis uses a series of self-learning algorithms to build a history of normal transactions. It can compare future events to that set of data to identify anomalies. Although FHWA and the researchers initially discussed sharing the information with real-time enforcement agents, they decided instead to hold the data collected for later analysis. It is somewhat common for a driver to mistakenly report a destination when leaving a terminal or to divert all or part of a load to another destination and report it correctly later. Thus, stopping the vehicle at this time would not be a good use of resources. Enforcement agencies could always implement action later if needed.

Early Lessons Learned

During the initial phase of the research, the Oak Ridge team worked to identify sensor equipment that would be resilient and able to conform to existing communications standards and equipment on tanker vehicles.

That early work included the development of fuel markers. The idea was to introduce nanomarkers into fuel so that when an operator was unloading it from a tank truck, a sensor mounted in a valve would identify the concentration of the marker. If the sensor reading was different than expected, it would signal that the fuel had been changed in some way. The Oak Ridge research team developed a robust marker and sensor that were able to work in extreme conditions.

Next, the team computed the break-even cost for employing such a sensor and marker. The research team computed the cost using consumption data from the U.S. Energy Information Administration. The team determined that the cost of developing and deploying the markers would be more than four times the break-even cost, which made it unlikely that the industry would deploy the markers. So the researchers suspended further work in this area.

They then began work on an evidential reasoning system that could “learn” the habits of the drivers. That system includes software on both the onboard devices and a back-office system, which could be located anywhere, but most likely at the operational headquarters of the carrier. The data onboard the vehicles would be communicated by several optional methods, including as cellular data, to the back-office system on a predetermined timeframe, such as at the end of the driver’s shift. The downloaded data then would be compared to historic activity to identify any questionable events.

As more data are entered into the back-office database, a more reliable normal set of events would be created to compare new events against.Although this would likely be done on an individual driver’s basis, there is no reason why a full fleet standard could not be established. For an event deemed to be suspect for fuel tax evasion (for example, destination stated at a fuel loading location different from the actual destination), the evidential reasoning system could create a report of the exceptions for further analysis by the company or auditors.

Delivery vehicles have two sets of valve releases. One is an emergency release, which the driver opens before the unloading starts. The operator then opens the primary valve. With a certain sequencing of these valves, however, a driver could unload small amounts of fuel, presumably outside the planned delivery. Using valve-sensing technology and evidential reasoning software, the system would identify this as an unexpected event, and log the time and place. If nothing else, this would serve as a deterrent for such activity.

The Oak Ridge researchers were able to find several partners for proof-of-concept testing of the various sensors, communication devices, and software they had developed. One of the partners was Pilot Flying J, which operates truck stops across the United States and a fleet of transport vehicles.

The proof-of-concept testing, however, did not lend itself to the collection of real-world data because the equipment and sensors were not certified for hazmat tanker use and were not environmentally hardened, as they showed signs of degrading when exposed to normal weather conditions. Further, to gather the required data, the testing steps and other maneuvers were difficult to control, because the research team would be unable to request tests of particular scenarios since fuel deliveries are done on a rigid schedule. These factors and general concern for staff safety and the safety of the Pilot Flying J test vehicle led to a decision early in the project to conduct this data collection at a closed test facility instead.

In November 2011, the research team issued Supply Chain-Based Solution to Prevent Fuel Tax Evasion: Proof of Concept Final Report summarizing the project’s first phase. The report is available at http://cta.ornl.gov/cta/Publications/Reports/ORNL_TM_2011_132.pdf. In the report, the team concludes that the solution is workable and provides information for further study. Although the original proposal was for the identification of actions to reduce fuel tax evasion, it became evident that such a system could also be used to combat fuel theft.

Moving to Real-World Testing

Based on the positive results from the first phase, FHWA approved additional work titled Safeguarding Truck Shipped Wholesale and Retail Fuels. During this second phase, the various sensors, communication protocols, and software integrations were to be tested under real-world conditions. Using lessons learned from the first phase, the team used existing communications on the vehicles, which provided less exposure of the sensors and wiring. Also, the expanded testing would allow for much greater analysis and potentially more varied scenarios.

In order to implement the real-world testing, the Oak Ridge team sought equipment partners. Barger Transport, which operates primarily in eastern Tennessee, Virginia, and Kentucky, provided three identical tanker vehicles. Other companies arranged additional support: Air-Weigh (wires and harnesses), LBT, Inc. (formerly Liquid & Bulk Tank) (trailers), and Innovative Software Engineering (onboard telematics devices).

The research team placed sensors on the loading/unloading valves at the base of the trailers and on the top hatches, plus onboard weight sensors to measure steer, drive, and trailer weights. The top hatches on tankers are used only for maintenance. If operators open those hatches during normal operations, that could signify suspicious activity. The sensors on the loading valves can be associated with physical location (through the use of GPS) and could help establish loading and unloading events. Finally, the vehicle weight sensors could combine detected changes with the other information to identify the time, location, and volume of scheduled and unscheduled events.

Drivers operated the three tanker vehicles in the pilot test for a period of 8 months. During that test period, the drivers filed nearly 700 fuel logs, each representing a day’s worth of deliveries for a single vehicle. The tankers delivered more than 7.5 million gallons of fuel during the pilot, and the drivers logged more than 375,000 miles (603,500 kilometers).

After loading the fuel on a delivery vehicle, the operator would enter the fuel type and quantity into a small telematics device in the truck cab. The fuel tax evasion system would require, at least in its proposed deployment, these additional tasks.

Today, telematics devices are widely used in the trucking industry to monitor certain systems, vehicle performance, and driver hours. They are generally the size and shape of the GPS tracking devices used in automobiles. Tanker vehicle drivers use telematics devices for additional purposes, such as to enter time-of-service information. Thus, the system as proposed could be integrated into existing devices. The drivers had to learn the new process but were able to do so with minimal issues.

The telematics devices will transmit the data to a back-office system maintained by the transport company. Like many computer systems today, the data storage could be in the company’s office or in some cloud-based storage. The transport company could query it in a number of ways, including operations by vehicle, by driver, and by anomalies, such as vehicles traveling different routes than expected. The system has the flexibility to capture and analyze the performance data in separate applications. For example, the carriers might collect data on the vehicle operations, while the fuel distributors might receive just the information on the actual fuel deliveries.

“The system developed in this project is basically transparent to the driver,” says Oscar Franzese, Ph.D., who led the work at Oak Ridge. “The operation of valves when loading and offloading fuel is the same with or without the system deployed. However, the information collected with an active system, which is fed to self-learning algorithms developed for the project, allows identifying normal driver operations as well as unusual activities. This is beneficial for carriers and for tax auditors.”

While some of the researchers developed hardware related to the sensors and the wired connectors, other members of the team worked on software development. The evidential reasoning system they developed uses a self-learning algorithm to process the data and establish probable sequences and timing of events that over time would be able to establish probable actions by drivers. When events exceed the parameters (for example, valves opened longer than normal), the system can create real-time notices, or save the event to be analyzed at a later date.

In addition to the on-vehicle implementation, the research team developed a small-scale reproduc--tion that enabled them to test the valve sequencing and the accuracy of the data shown on the telema-tics devices.

Data Collection And Analysis

The objective of the research was to build a system of integrated technologies that record all the activity of a fuel transport vehicle to facilitate the identification of any unusual events. Users would be able to compile the data and query it in a number of ways, including to validate activity reported for tax purposes.

To maximize the benefits of the system, FHWA empaneled a technical working group charged with identifying the type of information that they would like to see. The group’s members were fuel tax auditors and enforcement personnel: Al Howard, Internal Revenue Service (retired); Dawn Lietz, Nevada Department of Motor Vehicles; Rodney Pendley, Tennessee Department of Revenue; Wayne Rhoads, Alabama Department of Transportation; and Chuck Ulm, Field Enforcement Division of the Comptroller of Maryland.

The working group provided input into the development of the Fuel Distribution Auditing System (FDAS), which is an optional module of the overall system developed by Oak Ridge. The FDAS enables users to direct queries to the collected data.

For example, if an auditor or company employee were looking for all activity in a given State (or area such as a city), the researcher could enter the relevant parameters and create a report with that activity. If a delivery is reported as destined for one State, which is reflected on the bill of lading, but has to be diverted to another State, that diversion would need to be reported. Many States require the carrier to report the diversion to a central database. The auditors would be able to query the database for all deliveries to a certain State. The system would report the actual activity, as well as what the intended action was, so the auditors could identify mistakes in the reporting of the proper delivery location, plus incorrect tax payments.

“Verification of delivery locations has always been challenging,” says Pendley, one of the working group members and tax auditor supervisor for the Tennessee Department of Revenue. “Tennessee, sharing borders with eight other States, has always had issues with untaxed and reported fuel shipments entering the State. Access to the system database, which contains the geographic locations of fuel movements, will be a great tool for verifying deliveries and identifying questionable transactions.”

Next Steps

The research team did not design this long-term project to be a rigid application, where the user has little ability to customize settings or reports for their own operations. Rather, the research proved that an integrated system can monitor the activity of a fuel delivery vehicle and identify in real time, if desired, any activity that might suggest fuel tax evasion or even fuel theft. Although early efforts would have limited the application of the technology to fuels (through markers), the final conclusions show that such systems are applicable to any number of transported commodities. The findings suggest an overall system architecture but do not limit the possible alternatives. All of the components, including the software, would be scalable--benefitting small single truck operations all the way up to large fleets.

During the project, the team shared the potential of the system with industry, especially trailer and equipment manufacturers. The researchers also gave a series of presentations to the Federation of Tax Administrator’s Motor Fuel Tax Section, whose members include fuel wholesalers and distributors, as well as State auditors and fuel tax administrators. In addition, other Federal agencies showed interest in the work, including the Federal Motor Carrier Safety Administration, the Transportation Security Administration, and the U.S. Department of Defense.

The research team and the FHWA Office of Highway Policy Information will continue to share the results with entities that might wish to employ such a system and to whom it could be licensed.

David Winter, director of FHWA’s Office of Highway Policy Information, says, “This research has demonstrated that it is possible to create an integrated system that can help validate the movement of motor fuel, assure that the proper taxes have been paid, and give the distributor confidence that the proper delivery has been made. Although this will benefit the taxpayers and the fuel industry, we see that there are applications for other logistics operations as well.”

Michael Dougherty is a program analyst in FHWA’s Office of Highway Policy Information. In this role, he is responsible for the Highway Use Tax Evasion and Heavy Vehicle Use Tax Programs. He holds a bachelor of science degree from Towson (MD) University.

For more information, contact Michael Dougherty at 202–366–9234 or michael.dougherty@dot.gov.