On-highway tractor-trailer vehicles operate in a complex aerodynamic environment that includes influences of surrounding vehicles. Typical aerodynamic analyses and testing of single vehicles on a test track, in a wind tunnel, or in computational fluid dynamics (CFD) do not account for these real-world effects. However, it is possible with simulation and on-road testing to evaluate these aerodynamic interactions.

Multi-vehicle, on-road aerodynamic interactions have been the subject of physical testing and reporting worldwide since at least the 1990s, under the phrases “slipstreaming,” “platooning,” or “intelligent highway” research. These terms overshadow the fundamental truth, that vehicles share road infrastructure and generally are more likely to be operating in the wake of other vehicles than not. The correct term for this is generally defined as “traffic.”

As part of the U.S. Department of Energy SuperTruck program that was awarded to Cummins Inc. and Peterbilt Motors Co., a series of investigations was conducted on the aerodynamic ramifications of traffic. One such investigation, carried out by experts from Peterbilt and design software provider Exa Corp., focuses on the impact of traffic conditions on aerodynamic drag for typical vehicle separation distances at typical highway speeds. Testing was performed along two Texas highway routes to determine the impact of surrounding traffic conditions on oncoming airflow to a vehicle. Multi-vehicle CFD was also employed.

SuperTruck aerodynamics

The SuperTruck program for the Peterbilt/Cummins team challenged industry to develop near-production-ready technologies for attaining 50% improvement in freight efficiency over a representative road test cycle and 68% for a typical 24-h single driver operation relative to a MY2009 representative baseline vehicle. These aggressive goals required a systems approach to the vehicle design, with complementary contributions from many systems. The significant goal from aerodynamics was a 48% reduction in wind-averaged aerodynamic drag, as determined from CFD analysis in a single vehicle virtual wind tunnel.

Validation of the system-level performance for the road test cycle was accomplished through on-highway, two-vehicle-modified SAE Type III tests, over a predefined 310-mi (500-km) course as reported in business confidential information (CBI) SuperTruck status reports to the DOE. Validation of the 24-h cycle test was done with a combination of SAE Type III tests over a predefined 500-mi (805-km) course and overnight static hotel segments. Per the program requirements, vehicles were operated at 65,000 lb (29 t) gross vehicle weight.

During testing of the first prototype Peterbilt Demo 1 SuperTruck, the on-highway fuel economy results were found to be better than predictions from summing the contributions of the interacting systems. The evaluation suggested that the aerodynamic contributions were exceeding predictions, primarily from uncontrolled ambient conditions that include weather conditions and traffic that were omitted from the original CFD evaluations.

The relative magnitude of the weather effects were fairly well understood from various source documents on wind and temperature effects. The magnitude of traffic effects on drag prediction was not as well understood. An on-highway test was developed by Peterbilt engineers, using the two SuperTruck tractor-trailer units on a straight, relatively flat stretch of divided highway operating at posted highway speeds. This testing characterized potential magnitudes of airflow changes due to separation distances. An additional blocking vehicle was employed to eliminate random interactions from other vehicles.

Physical testing

A challenge was how to measure the wake effects on the trailing vehicle without modifying the vehicle with additional instrumentation that could bias results. Peterbilt engineers observed that the engine fan was effectively operating as a simple flow meter as it free-wheeled with the clutch disengaged. Instrumentation was added to the fan to measure the rotational speed, which augmented the continuous data recording. In flat, on-highway operation, the cooling fan is generally not needed, with cooling provided solely by ram air.

The MY2009 baseline Peterbilt Model 386, pulling a standard 53-ft Utility DX4000 van trailer, was selected as the lead vehicle. The trailing vehicle was a Peterbilt Model 587 based tractor, pulling the Peterbilt-modified Demo 1 SuperTrailer. Both had production engines and cooling systems. The 587 tractor-trailer system is a more aerodynamic configuration, with added aerodynamic features such as tractor-trailer gap closeouts, trailer skirts, and tractor bogey aerodynamic shields. The two vehicles were operated with vehicle spacing reducing from multiple vehicle lengths down to 10 ft (3 m). Multiple tests were conducted in opposite directions on the divided highway.

This physical testing established that the free-wheeling fan rpm reduced as separation distances decreased. Fan speed was measured using a Banner SM312LV optical sensor. The validity of the optical sensor was checked when the fan was engaged. The results from the optical sensor showed the engaged fan speed within 1% of the engine speed multiplied by the known fan pulley ratio of the engine driven fan.

Subsequent review of the 310-mi on-highway test runs and evaluations on an alternate commercial 400-mi (644-km) run from a warehouse in Irving, TX, to a Laredo terminal showed that free-wheel fan speeds could document the presence of multi-vehicle interactions. An rpm change threshold was set from open free stream on-road conditions. The full route data was then parsed; significant rpm reductions due to surrounding traffic conditions occurred between 37% and 53% of the travel.

In this testing, full vehicle aerodynamic data was not collected either through fuel-economy (FE) measurements or other means. However, recent SAE Type 2 FE testing has shown that running tractor-trailer combinations in proximity can achieve FE savings through aerodynamic effects on both the leading and trailing vehicle.

Having substantiated that wake vehicle interactions were occurring with sufficient frequency and magnitude to bias predictions, a detailed CFD investigation was initiated.

Multi-vehicle drag CFD evaluations

Simulations using PowerFLOW, a CFD software based on the Lattice-Boltzmann Method, were used to evaluate the aerodynamic drag impact on leading and trailing tractor-trailer configurations in a drafting configuration at various trailing distances. PowerFLOW is an inherently transient solver that utilizes a Very Large Eddy Simulation technique where the large anisotropic levels of turbulence are calculated on the computational grid while the small isotropic scales of turbulence are modeled using a two equation turbulence model based on extended Renormalization Group theory. This simulation technique has developed best practices that have been used to validate against simplified and detailed heavy truck configurations in wind tunnels.

These best practices have been applied to both the leading and trailing vehicle in the simulations presented. The simulation methodology also conforms to SAE J2966, which provides guidelines for aerodynamic assessments of heavy vehicles. The fluid domain between the leading and trailing vehicle in the simulations was maintained at the same resolution used to characterize the wake of each individual vehicle. This was deemed appropriate for this investigation, but further grid resolution studies could be performed in the future.

The CFD analysis was conducted on models with fully detailed external flow surfaces and underhood surfaces of the two tractor-trailer units. The geometry was obtained from original CAD manufacturing models and refined from an intensive photogrammetry scan of the actual vehicles to capture the as-built configuration as accurately as feasible. The leading vehicle in the CFD simulations was a detailed representation of a Peterbilt 386 leading vehicle in the on-road testing; the trailing vehicle was a later version of the SuperTruck based on a Peterbilt 579 platform and was chosen for simulation for SuperTruck program considerations. Since exact correlation is not desired for this study, the differences in the models were deemed acceptable and will highlight the overall effects vehicles experience in a drafting configuration. Both vehicles were modeled with rotating tires as specified by a rotating wall boundary condition in conjunction with a rolling road floor condition and in an open road or very low blockage environment.

Drafting simulations were completed at separation distances of 30, 60, 100, and 160 ft (9, 18, 30.5, and 49 m). No offset in the direction perpendicular to travel was used. Separate simulations at 0° and 6° yaw angles were completed for each of the separation distances at free-stream velocities equal to 60 mph (97 km/h). Simulations with the leading and trailing tractor-trailers by themselves were also performed and provide the values for normalization of aerodynamic performance without any traffic effects.

At a 30-ft separation distance, the yaw averaged drag of the leading vehicle is reduced by 7% compared to the same tractor-trailer without any traffic flow disturbances. Analysis of the cumulative drag along the vehicle shows that most of the drag reduction for the leading vehicle occurs at the tail end of the trailer. The stagnation pressure from the trailing vehicle communicates with the wake pressure of the leading vehicle, causing an increased base pressure on the leading vehicle and an overall drag reduction.

The increase in pressure on the back face of the leading tractor-trailer configuration reduces significantly between 30 and 60 ft.

The drag reduction for the trailing vehicle is shown to be more significant than on the leading vehicle. At a 30-ft separation distance, the drag on the trailing tractor-trailer is reduced by 17%. The main driver for the reduction in drag is the reduction in oncoming velocity to the trailing vehicle from the wake of the leading vehicle due to the presence of the leading vehicle wake. The velocity upstream of the no-traffic condition is as expected near free stream with the only impact being the stagnation zone in front of the tractor.

With a 30-ft trailing distance, the oncoming flow onto the trailing vehicle is generally at 50% of the free-stream velocity. On the surfaces of the vehicle, this reduces the stagnation areas like the hood, bumper, and windshield, but also reduces the suction pressure on curved surfaces designed to accelerate the flow like the roof fairing. The flow on to the trailing vehicle at 60, 100 and 160 ft increases with increasing separation distance and is not uniform.

At 60-ft separation distance, the velocity near the center of the vehicle, where most stagnation zones are, is lower than the velocity at the outer surfaces, where more flow acceleration and suction pressure is obtained. This is the driving factor for the slight reduction in drag for the trailing vehicle as the separation distance increases from 30 to 60 ft.

For the trailing vehicle, the aerodynamic drag is not the only technical area needed for investigation. The lower velocity in front of the trailing vehicle also impacts the amount of cooling airflow through the underhood because of the change in the driving stagnation pressure. If not enough airflow is driven from the ram air effect, the fan clutch will engage to drive the fan necessary to ensure proper operation of heat exchangers critical for powertrain and air-conditioning requirements. The fan operation draws power from the powertrain and can impact vehicle FE as a result.

The findings in these simulations show that running in a drafting or platooning configuration can have a drag reduction and thus a FE improvement. Communicative braking systems between vehicles and changes to safety regulations may be needed to realize this drag improvement on the road. Drag savings for the separation distance of 30 ft are significant, and the analysis shows a total drag savings of 23% are possible. At highway speeds, this could translate to 11.5% FE savings for a two-truck platooning configuration. Recent on-road FE testing has shown similar findings for lead and trailing tractor-trailer efficiency.

This study highlights that real-world heavy-truck aerodynamic prediction through test and analysis methods can be improved by incorporation of significant environmental factors such as traffic conditions.

This article is based on SAE International technical paper 2014-01-2436 by Jeff Smith and Rick Mihelic of Peterbilt, and Brandon Gifford and Matthew Ellis of Exa Corp. Their paper is part of the “Vehicle Aerodynamics” technical session at the SAE 2014 Commercial Vehicle Engineering Congress, taking place October 7-9 in Rosemont, IL (http://www.sae.org/events/cve/).