The Traffic Accident Reconstruction Origin -Article-

[Home] [ARnews] [Contents] [Classified] [Advertisers] [Approach Angles] [E-mail Directory] [Feedback] [Organizations] [Reference Library]

Understanding The Vericom VC-2000 Accelerometer- Pitfalls for the Operator and Making Comparisons to Other Drag Factor Determination Methods

by

Ed Livesay

ACTAR reg#595

Note: This article is based upon excerpts from "DESIGN, CREATION, AND PROPER USE OF A DRAG DEVICE FOR THE DETERMINATION OF DRAG FACTOR," a publication tentatively approved for publication by IPTM in 1998.

The Vericom Corporation's VC-2000 Performance Computer presents a viable method of determining the skid-resistance of a given roadway surface. Like any instrument, the VC-2000 can be abused or misunderstood, resulting in erroneous values for drag factor. The key to avoiding such error is to understand the instrument's abilities and limitations.

The VC-2000… a "Capacitance Accelerometer"

Photo courtesy of the Vericom Computers, Inc., 6008 Culligan Way, Minnetonka, Minnesota 55345

In order to understand the VC-2000 we must first come to an agreement as to what we mean when using the word "Drag Factor" and other related terms. Therefore, for the purposes of this discussion and the use of the instrument, the following definitions will be used:

"Drag Factor" is a measure of the dynamic friction generated between a vehicle tire and the road surface of interest. It is the coefficient of friction between these two surfaces (often referred to as "mu"). Because this parameter is slightly dependant upon sliding velocity, it may or may not accurately describe the actual deceleration of a skidding vehicle (Delta Velocity / time). This is often the value found with drag devices which are pulled parallel with the test surface.

(The classic friction models offered by most educational physics texts are often erroneously expanded to include the behavior of a vehicular skidding action. Most of those texts plainly warn that Coulomb's rules regarding velocity and surface-a rea independence apply only to non-malleable, hard surfaces.... a vehicle tire is malleable and can conform to the macrotexture (irregularities) of a road surface. This causes slight velocity and surface area dependence... simple fri ction models tell only part of the story. Simple "Drag Factor" relates only one of many values for friction encountered by a vehicle skidding from one velocity to another.... A single value does not accurately describe a skidding vehicle's dece leration (unless that single number represents average drag factor or Delta Velocity / time).)

Drag Factor will not include other frictional or deceleration events such as atypical braking, mechanical delay/spin-down, grade, or slowing with an ABS equipped car.

"Skidding" means a full, locked-skid for all tires. Brakes must be in good repair making all tires skid for a relatively similar distance when applied. Simply stated, skidding means that the car is moving and the tires are not rolling.

"Ordinary Tires" have Shore A Durometer values ranging from ~55 to ~75 as tested per ASTM 2240-97. Contrary to popular belief, large truck tires more often than not conform to these same specifications. However, for other reasons, large trucks develop less skidding resistance in relation to passenger cars and light pickups, etc.

"Average Drag Factor" is one of several values reported by VC-2000s and velocimeters as well as drag devices pulled at their "proper pull angles." It is an actual skidding vehicle's Delta Velocity / time. The skidding vehic le is assumed to replicate the skidding characteristics of a typical passenger car with properly operating brakes. Depending upon how a VC-2000 is activated (auto threshold or brake-light connection), this value may include both mechanical delay and spin -down. This is the value that reconstructionists most often employ within skid and other speed-determination formulae when Effective Drag Factor is not available.

"Effective Drag Factor" is the average drag factor for a specific vehicle. This value might be generated with a vehicle with less than optimal braking, a vehicle that is not a passenger car (large truck or motorcycle), or a veh icle which in some other way does not replicate a typical passenger car with standard brakes. A vehicle equipped with Anti-Lock brakes would be an example. This value is probably the most accurate in describing the deceleration rate for a vehicle.

"Acceleration" will be used in its true sense: It is Change in Velocity divided by Change in Time. To the layperson's consternation, acceleration isn't limited to describing how fast something "takes off," but can also describe how fast something "slows down." For the knowledgeable, the term "accelerate" simply means to change the speed of an object (in either direction).

"Mechanical Delay"- The most recent definition of mechanical delay refers to the phase of brake application beginning with pedal depression and ending with any resulting change in wheel rotational velocity. The vehicle does no t appreciably slow during this phase which lasts only about a tenth of a second or so. No tire marks are created during Mechanical Delay. (Older definitions of mechanical delay initiated with pedal depression and ended at complete four-wheel lockup and typically lasted from about two to 4-tenths of a second.)

"Spin-Down" begins at the end of mechanical delay (as previously defined) and ends with total four-wheel lockup. This phase can last about 3-tenths of a second and includes tires marks ranging from light to dark in appearance. At the point at which the vehicle's forward velocity differs from a wheel's rotational velocity by about 20%, "slip-friction" is encountered and presents the maximal values for friction observed throughout the skid. Some refer to this value as Peak Drag Factor. Development of Peak Drag Factor, (slip-friction) is the basis of ABS braking. More often than not, a VC-2000's default autostart threshold initiation settings cause the device to activate at the beginning of spin-down (whe re an experienced investigator would detect the beginnings of "shadow," "incipient," or "impending" skid).

So in summary we have a road-surface's Drag Factor, the Average Drag Factor for average vehicles, and a particular vehicle's Effective Drag Factor.

Now, with basic definitions in hand, we must come to understand how the VC-2000 works and what it measures.

VC-2000 Description and How it Works

A VC-2000 is a "single-axis, capacitance type accelerometer." The device measures acceleration, in terms of "G-forces" in a single direction. At the heart of an VC-2000 is a small mass which is held in position between two small blocks or walls of metal with a spring on either side of the mass. The walls are solidly attached to a circuit board, which is attached to the accelerometer's housing. Finally, the whole mess is solidly attached to the test vehicle. One of the tiny walls of metal is located towards the front of the test vehicle while the other wall is located towards its rear. The mass and its mooring springs are found between the two walls.

The "Idea Behind a Capacitance Accelerometer"

While motionless, or while traveling at a constant velocity, the mass is evenly spaced between the two capacitor walls. Newton's 1st Law of Motion controls the mass. It attempts to "remain in state," moving or stationary. When the test vehicle slows or speeds up (accelerates), one spring is compressed and the other is extended, both acting to change the velocity of the mass. As a result, the mass is closer to one of the metal capacitor walls while accelerating. The greater the acceleration, the smaller the distance between the mass and one of the walls. The distance between the mass and either the front or rear wall is electronically measured 100 times per second (in a VC-2000). The unit's electronics actually measure the capacitance (an electrical property), which is related to the distance between the mass and the walls. (For accuracy's sake I must reveal that the transducer within the VC-2000 doesn't really use coil springs or metallic walls. It actually employs a mass held in place by a piece of tin which bends like a leaf spring. The coil spring description is simply easier to visualize by some). The transducer is actually a small, 3-layer sandwich-like component.

The spring is very carefully chosen. Too much spring stiffness causes the movement of the mass to lag while too little stiffness causes a bounce-like effect. This whole movement-control process is termed "dampening" and is quite sensitive. The choice of dampening parameters greatly affects accuracy and has differentiated the various brands of accelerometers in the past. The VC-2000 is "critically" damped, resulting in a very accurate compromise.

The other essential component if the VC-2000 is an accurate electronic/crystal clock.

Graduates of engineering curriculums or of the typical "Advanced Accident Investigation" course can appreciate what can be calculated with time and acceleration measurements. Algebraic manipulation can reveal a test vehicle's speed, translation (distance) while accelerating, applied horsepower, etc. More on this idea will follow.

VC-2000 Setup and Use

In order for the VC-2000 to report a drag factor it must be attached to a car which is then used in an actual skid. All the usual procedures for skid testing apply. For maximal accuracy, use a similar car, in the same direction of travel, in the area of interest. A skid test employing a VC-2000 is quite easy to accomplish, but proper planning fo r traffic and other safety issues is always important.

Checking the calibration of a VC-2000 prior to use is not mandated but is probably prudent to assure the ability to testify to its accuracy. Checking the calibration is quite simple and very similar to the process of checking the calibration of a R.A.D .A.R. unit with a tuning fork. The "tuning fork" used to check an accelerometer is the acceleration of gravity. A user simply starts the unit in "continuous" measurement status while the unit is stationary upon a flat surface. The unit is then placed upon its front face (hanging off the edge of a tabletop so that the readout screen is still partially visible). When properly calibrated, the activated unit will reveal +1.00 "g" (with a very small acceptable variation that depends upon the elevation and location upon the earth's surface). When placed upon its rear, the unit should reveal -1.00 "g" (note the sign change).

Some choices must be made concerning how a user wishes his/her VC-2000 to be activated. "Activation" refers to how the VC-2000 determines that a skid event has begun. This internal calculation process can be started in any of three fashions.

The first method is to employ the unit's "Christmas Tree" mode. This method mimics a drag strip's start-light sequence "tree." The proper button-pushing sequence causes the device's readout to display sequential numbers like a countdown. The operator watches the countdown and "launches" the vehicle at the sequence's end. In this mode the VC-2000 will even reveal when the operator "jumps the gun" and moves the vehicle prematurely. (The VC-2000 is used by many drag-strip racing teams to tweak maximum performance out of man and machine… that's why the device's label reads " VC-2000 Performance Computer.")

A second method employs an electrical hook-up with a vehicle's brake-light system. All but the basic model VC-2000 have this option. A user electronically connects an optional wiring harness leading from a special socket on the rear of a VC-2000, to an appropriate connection point in the brake-light system. With this wiring harness in place, the device will automatically start its calculation process upon receiving a 12-volt signal from the brake lights or upon reaching the preset Autostart G-force threshold (whichever occurs first). This option causes a VC-2000 to record and report not only spin-down, but mechanical delay as well.

The third and last start method for a VC-2000 is to make use of its internal "Autostart G threshold." The "out of the box" default settings will cause the unit to start its calculation process when the accelerometer measures 2-tenths of a g (either slowing or starting). This setting will not-so-coincidentally start the unit's calculation process at about the point at which a trained investigator would find the beginning of a skid mark (the shadow, incipient, or spin-down portion observed some appreciable distance prior to clearly visible, dark skid mark). Since this point is found after the vehicle has slowed to some slight degree, the VC-2000's calculated value for Average Drag Factor will probably be a bit higher than a velocimeter (i.e. the MacInnis Fifth-Wheel) or another VC-2000 started via brake-light hookup. The VC-2000 Braking Test Computer Owner's Manual states "When comparing the VC2000 to any other device, the two devices must start at exactly the same time for the data to be meaningful."

After the initiation mode is selected the VC-2000 must sometimes be "tuned" for a particular test vehicle. This is done through the adjustment of two settings. Firstly, the Autostart G threshold can be adjusted. This is usually done to get better accuracy when measuring a vehicle's take-off ability than when determining skid-resistance. However, it might be necessary in testing the braking ability of a vehicle with very poor brakes, when testing very slippery surfaces, etc.. It is obviously necessary when acceleration is less than the .2g default threshold as might be the case when evaluating the skid resistance of an icy roadway.

Secondly, a VC-2000's default setting for pitch calibration (referred to as "CAL Factor" in the manual) might need to be abandoned. Recall that a VC-2000 is a single-axis accelerometer. This means that it measures acceleration in one straight line or axis. When a user sets up a VC-2000 for a braking run, the unit must be installed within 5-degrees of level within the test vehicle. The unit is capable of sensing and compensating for slope caused by minor level error. However, when a skid begins, " weight-shift" occurs and the nose of the test vehicle dips and its rear rises. The designer's of the VC-2000 wisely accounted for this within the unit's default pitch calibration setting (No. 1 for most production vehicles). Unusually stiff suspension systems or long wheel-bases might require lowering the setting to No. 0. Very "squishy" suspension might mandate the use of pitch calibration setting No. 4. Some degradation of accuracy occurs with the higher settings according to the manufact urer.

Close "matches" of accurate speedometer readings and VC-2000 calculated speeds during tests is a sign of proper settings for G threshold and pitch calibration. They measurements don't match perfectly every time, but they should be relatively close.

The problem presented by weight-shift during a skid is repeated (at 90 degrees) in cases where the test-vehicle yaws to any appreciable degree. Remember the VC-2000 is a single-axis accelerometer. The axis measured extends directly ahead and to the rear of the test-vehicle. The greater the yaw angle the less the line or axis points in the direction of acceleration. In the extreme, a VC-2000 mounted in car that had yawed 90 degrees would not sense any acceleration at all. Just as in pitch, a slight yaw introduces some degree of "sine-error." As previously mentioned the unit can handle small variations, but significant angles cause the device to report erroneous values. I recall a story about a motivated skid-tester (names aren't really necessary). He wished to determine the skid-resistance of an icy road-top. Without much forethought, he proceeded to slam on the brakes with his new and trusty VC-2000 ready to go. His vehicle's rear tires locked first and the car began to swap ends. Needless to say, the "pucker factor" vastly exceeded the drag factor before he finally came to a stop. Gross yaw occurred and the VC-2000's calculation process ended prematurely. The reported average drag factor was much higher than reality allows atop ice.

VC-2000 Background Information and Operational Considerations

It is important to understand that a VC-2000 does not measure speed or distance directly. It measures acceleration and it measures time. It then uses these parameters to calculate distance. This accomplishment is relatively straightforward. Determining values for acceleration 100 times per second is, in itself, an amazing feat. But, knowing the acceleration history of a skid and its time allows the VC-2000 to calculate the velocity change. A simple example will clarify.

Drop a pencil from desk height. What is its acceleration? Of course it accelerates toward the ground at 1g (32.2 feet-per-second-per-second or 9.8 meters-per-second-per-second). Like the VC-2000, we know the acceleration of the falling pencil. But, with the additional information of how long (time) the pencil falls, we can calculate its speed with the relationship

In a similar way, the VC-2000 calculates the changing velocity profile of the skidding car from its changing acceleration. This how a device measuring only acceleration can "know" what speed change also occurred. It is calculated!

Now the VC-2000 uses the calculated velocity history and time to calculate the distance skidded. Again, distance is calculated. Don't expect a VC-2000's distance measurement to closely match the average visible skid length… it simply won't happen often. However, the unit's value for distance might be closer to what the investigator might consider to be "accurate" than measurement of visibly detected skids.

Mechanical Delay & Spin-Down can also have a significant effect on skid testing. Remember from the definitions that Mechanical Delay plus Spin-Down present the time beginning with the foot's first contact with the brake pedal (Brake Application) until the tires actually lockup.

The effects of mechanical delay causes differences in values for f from various measurement devices & methods.

The period of time beginning with brake pedal depression and ending with full, four-wheel lockup includes several actions. Mechanical Delay includes the time required to employ the brake system's hydraulic components to deliver sufficient force to aff ect the rotational velocity of the wheels. Spin-Down adds the time to overcome the wheel's spinning inertia as well as the friction developed until the wheels actually lock. Aspects ranging from brake design efficiency to roadway drag factor can all eff ect the time required to complete the Mechanical Delay and Spin-Down phases of a skidding event.

Mechanical delay, Spin-Down, and the accelerometer-activation method chosen are usually the culprits that cause the largest differences among the various methods used to quantify Average Drag Factor.

In some studies (i.e. SAE# 930665, Jerry Eubanks, Rusty Haight, et.al.) the default Autostart Threshold settings cause the VC-2000 values for deceleration to better match values calculated by physically measuring visibly detected skid distances. In short, the VC-2000 better agrees sometimes with good, old-fashioned skid tests, not because it's more "accurate," but because it happens to measure things in a way that more closely mimics how investigators measure things with Rolotapes and tape-measures. Other devices, or VC-2000s started with brake-light hookups, will report greater skid distances and lower Average Drag Factors because they start at brake pedal depression instead of at the beginning of Spin-Down.

Another consideration for accelerometers is resolution. The VC-2000 electronically measures acceleration every 100th of a second. The values are then lumped into groups of ten and averaged. The unit's screen and internal programming report those averaged sums to reflect a value for acceleration every tenth of a second. The more expensive models of the VC-2000 have computer ports and special software which allow the downloading of the 100th-of-a-second values for viewing and manipulating within your separate personal computer.

The user needs to recognize that a hundredth of a second seems quite small and serves well when your interest lies in racing, brake system design, or skid-resistance determination. However, there are occasions when a much finer resolution like a thousandth-of-a-second are mandated. An example would be an in-depth study of a vehicular collision. These events often last only 1 to 2 tenths-of-a second. Having only a standard VC-2000's one or two readout values wouldn't suffice. You'd probably prefer to have the much finer resolution provided by units which offer 1000th-of-a-second or smaller increments.

To illustrate the resolution, let's pretend that the geometry teacher has asked us to draw a circle upon a piece of paper. However, the rules of her game allow us to use only straight lines to draw the circle. When she allows us to use three lines or four lines, we can come up with a triangle and a square. Neither are circles. But, as she allows us to use more and more shorter straight lines, the image upon our paper begins more and more to look something like a circle.

Polygons with increasing number of sides- "A circle is a polygon with an infinite number of sides"

In a similar way, the more frequent the measurements, the better an accelerometer can accurately describe acceleration. This concept is the crux of Calculus over Algebra. A VC-2000 uses such a mathematical process to numerically integrate its many measurements.

Another limitation with the VC-2000 is its maximum 5-g measurement capacity. While this is plenty enough for racing and skid-resistance, this range is inadequate for recording most actual vehicle collision accelerations. These events usually involve quite a few more than 5 gs. A standard VC-2000 just wouldn't do.

It's important to realize that an accelerometer is an electronic instrument and can potentially be affected by temperature. Cooling typically causes shrinkage and heating typically causes expansion. Both phenomena don't necessarily occur consistently or proportionally. Since the important component in a capacitance accelerometer indirectly measures small movements over distance, temperature must be considered. The VC-2000 does employ a compensation scheme during its calibration process but, even though the manual does not specify an operating range, such a range probably does exist. I would suggest using caution in extreme conditions. It's probably a good idea to tape a piece of aluminum foil over the top surface of a VC-2000 so that it doesn't get too hot while perched within the windshield of a vehicle on a summer's day.

It is evident that in addition to possession of a VC-2000, one must possess a clear understanding of what the instrument is and is not revealing. Some vehicular acceleration problems can be more complex than one might think. In such cases the user must able to properly interpret an accelerometer's values for acceleration.

There are scenarios in which the skid-resistance value from a VC-2000 employed with default values might not be appropriate. An example would be a case in which the investigator needed a skid value for post-impact skidding of a vehicle (i.e., determining S3 in a momentum problem). The out-of-the-box value for Autostart threshold is two-tenths of a G. Initiating the device at a .20-g point would include a bit of Spin-Down. But, there is no braking application involved in this case (the driver didn't depress the brake pedal… the car was simply pushed sideways away from the impact area)! Wouldn't it be more accurate to set the Autostart threshold to a much higher value and thereby delete any mechanical delay/spin-down considerations and produce a value reflecting pure (but constantly changing) coefficient of friction? (This example is a bit more rhetorical than realistic… the difference in the two values would probably be insignificant…. But the point remains.) It might be that a properly used drag device would suit this case better.

The reader should begin to have an understanding why differing devices for skid resistance testing yields differing results. A properly designed, properly pulled, and "tuned" drag sled can yield values for Average Drag Factor that converge quite well with the VC-2000 values for Average Drag Factor. A VC-2000 properly set up for the test vehicle and activated by autostart threshold will provide Average Drag Factor (Spin-Down with varying Dr ag Factor) for a specific vehicle and roadway surface. Velocimeters (5th Wheels) and VC-2000s employing brake-light attachments will provide Effective Drag Factor including total mechanical delay and spin-down plus varying drag factor.< /P>

The MacInnis "Fifth-Wheel" (a "velocimeter")

(Photo courtesy of MacInnis Engineering Associates, #11-11151 Horseshoe Way, Richmond, BC, V7A 4S5, Canada, 1-800-565-3040 )

If this sounds confusing, be soothed by the fact that the differences in the several methods and devices, and their respective values for skid resistance are typically within .10 g of each other. In terms of speed, this typically translates to less than 5 miles-per-hour during most scenarios.

A properly designed and employed Drag Device being used by the author.

CONCLUSION

There are many different types of roadway friction testing devices. For general applications, the VC-2000 might be considered to be one of the better devices. The ideal scenario is one in which a VC-2000 is employed within THE vehicle involved in the collision, in THE area of the acceleration, and under THE same conditions. Experienced accident reconstructionists know that such opportunities are rare indeed. More likely, a user will probably employ his accelerometer within a different vehicle during test skids. Such a practice will suffice quite well as long as the user is knowledgeable enough to compensate for any significant differences.

It should be apparent that a g-triggered Autostart VC-2000 test can differ slightly from other accelerometers, VC-2000PCs, G-Analysts, and bumper-guns. As discussed, the method by which the particular device is activated is the primary key- just what information is being recorded? Devices that are activated by brake-light electronics will record a wee-bit longer time, resulting in slightly higher calculated velocities, and consequently slightly longer calculated distances resulting in Average Drag Factor values which are a little lower than a standard VC-2000. Old-fashioned, instrument-less test skids, which are measured visibly, will probably yield values that are considerably higher than those indicated by a VC-2000. A properly "tuned" drag sled, pulled at the proper pull angle, modified properly for inclines, and properly used will probably yield a value just a little lower than a standard VC-2000. An improperly set & used drag sled will yield values which are grossly higher than any of the other methods. With the exception of the improperly used drag sled, all of these instruments, including a standard VC-2000 will yield values which are within about a tenth of a g.

Ed Livesay graduated as valedictorian of Class 004 at North Carolina's Justice Academy (N.C.'s Police Academy) and has earned an undergraduate degree in Police Science and Investigations. Livesay has accrued over 2000 hours of investigation-specific course attendance. He completed the reconstruction curriculum from Central Missouri State University's National Police Institute in 1985 and has attended numerous additional courses at North Florida State University's I.P.T.M. and other similar entities. In 1994 Livesay was accredited by the Accreditation Commission for Traffic Accident Reconstruction and is a member of the National Association of Professional Accident Reconstruction Specialists.

After serving North Carolina's criminal justice system for 16 years, Livesay started his own licensed accident reconstruction service in Sanford, N.C., and has been accepted as an expert witness in local, State, and Federal courts. Since 1985, Livesay has taught in excess of 150 accident reconstruction and investigation courses to N.C.'s law-enforcement and other agencies including the NC State Highway Patrol, the NC Department of Transportation, the NC Division of Motor Vehicles, and several of North Carolina's Community Colleges. Livesay continues his career as Project Investigator and Accident Reconstruction Specialist for Almes & Associates, Inc. - Investigative Engineers, which currently serves the eastern U.S.

He can be contacted at elivesay@alltel.net

[Home] [ARnews] [Contents] [Classified] [Advertisers] [Approach Angles] [E-mail Directory] [Feedback] [Organizations] [Reference Library]