CAUSES OF A Presentation Athens, Greece · vehicles operated in the United States; however, the essential theory is valid in any typical configurati.on. Brake Operation A typical

CAUSES OF TRACTOR SHU-TRAILER ACCIDENTS

IN WET WEATHER

A Presentation for

The Second International Symposium on

Transportation Safety

Athens, Greece

The International Institute for

Safety in Transportation

by

Don L. Ivey, Richard D. Tonda, Walter B. Horne and Ted Chira-Chavala

from

Texas Transportation Institute Texas A&M University Sys~em

College Station, Texas

June 1985

ABSTRACT

CAUSES OF TRACTOR SEMI-TRAILER

ACCIDENTS IN WET WEATHER

by

Don L. Ivey, Richard D. Tonda, Walter B. Horne and Ted Chira-Chavala

Texas Transportation Institute

The discovery that heavy truck tires ao hydroplane at vehicle speeds from 50 to 70 mph is explored in depth. horne's prediction of this phenomenon is described in detail. The Texas Transportation Institute's testing program to verify this prediction is described and the results are compared with Horne's theory. An analysis of the Bureau of Motor Carrier Safety files on truck accidents for the years 1979 through 1981 shows the extreme overrepresentation of unloaded tractor semi-trailers in wet weather accidents, supporting the thesis that tire hydroplaning of large unloaded vehicles is a major contribution to wet weather accidents. Finally, other elements of the problem are explored such as multi-bottom vehicle stability, braking system effects, low tire-pavement friction and the speed increases associated with unloaded vehicles.

INTRODUCTION

1. The recent prediction by Horne (l) and verification (2) that truck tires are subject to dynamic hydroplaning at highway speeds has dictated a reassessment of the causes of tractor semi-trailer accidents in wet weather. Those concerned with highway safety have long noted the high frequency of single vehicle losses of control in wet weather. While many reports have been published documenting this phenomenon and discussing the influence of factors such as low tire pavement friction {3,4) reduced visibility (5) and hydroplaning {6) as influential factors, almost all this work has dealt primarily with automobiles. Large trucks have been considered a_Spe~ial and especially puzzling case. For example, some engineers have noted what seemed to be a comparatively frequent occurrence, the control loss of unloaded trucks during wet weather. Since it was understood that large truck · hydroplaning at highway speeds did not occur, these control losses were usually attributed to low tire-pavement friction, brake balance problems and possibly exces~ive speed by lightly loaded trucks. This common understanding has now been shown to be in error and another influential factor must be added to the list, truck tire hydroplaning.

2. The addition of this new factor does not negate any of those previously known but it does complicate the loss of control phenomenon since there is now a new level of interactions between 1) visibility,

1

2) speed, 3) loading condition, 4) pavement surface properties, 5) ·brake system characteristics, and 6) hydroplaning.

3. The reduction of visibility during wet weather has been described in several reports (5,7) and is generally associated with a reduction in the time available to make defensive maneuvers. This results in increasing the demand for traction performance at the same time the availability of traction for either cornering or braking may be greatly reduced. Figures 1 and 2 illustrate the influence of rainfall on visibility. Figure 3 (8) shows how available tire-pavement friction is reduced on wet pavement. This reduction is maximum at high speed.

4. Although it remains to be proven, it seems probable there is an inverse relationship between tractor semi-trailer speed and load carried. The obvious capability of some rigs to run faster under lightly loaded conditions points to this probability. It will be shown in other sections of this paper that it is the lightly loaded truck tire that is susceptible to hydroplaning. Thus the likelihood of higher speeds under reduced load condition increases the probability of hydroplaning, as well as making the appropriate balance in the brake system more critical to vehicle control.

5. It is these last two subjects, brake systems and hydroplaning that the next two sections will develop. Following those developments a recent analysis of heavy truck accident data will be presented. It will be seen this accident data is extremely supportive of other conclusions concerning the causes of truck losses of control in wet weather.

BRAKING, HANDLING, AND STABILITY

Before examining the details of this paper's thesis, it is instructive to review the essentials of braking and handling characteristics of tractor semi-trailer combinations, especially as they vary between loaded and unloaded conditions. This discussion centers on pneumatically actuated brakes as typically installed on commercial vehicles operated in the United States; however, the essential theory is valid in any typical configurati.on.

Brake Operation

A typical air brake system is shown in Figure 4 (10). The engine-driven compressor builds up and maintains pressure to the desired levels in the supply and service reservoirs. When the treadle valve is opened by the driver's foot (under normal operation) it takes a certain ·length of time for the supply pressure to be achieved in the brake chambers. This response time depends on the system's arrangement, which can be fairly complicated as·demonstrated in Figure 4. The volumes of each system component, orifice and valve sizes, and line lengths strongly influence the response time. Figure 5 shows time histories of trailer brake chamber pressure of three combination vehicles. The application times are measured from the instant of pressure increase at the treadle valve. This figure shows that, in an accident environment where a great deal can occur in a few milliseconds, the response time of the trailer brakes must have a significant effect.

2

Figure 1. Progressive influence of rainfall on visibility as the intensity increases and as the windshield surface becomes covered with water.

3

Figure 2. Spray generated by a tractor semi-trailer reduces visibility for drivers of following vehicles.

4

"'

150

; 1£ Ill CD 100 lli

i !i ., 50

I 0

lORY I

CRITICAL SLIP

I )WET I I I

25 50 75 100 TIRE SLIP , S , ,_

w 1 (w

Operating in the Brake Slip Mode (After Kummer and Meyer (3))

z 80 ., 0

o{ ::: 60 lli ::> z ~ 40 .J ., "' z ii: 20 Ill z

"' 0 0 0

0 4

c SELF -ALIGNING

TORQUE

DRY

I

I I

w'ET-

I I I ·rr 6 12 16 20

SLIP ANGLE 0( 0

~'!.j LTIRE PLANE

DRAG FORCE

Operating in the Cornering Slip Mode (After Kummer and Meyer (3))

Figure 3. Frictional characteristics of tires operating on typical wet and dry pavements.

~ .. '

I ! ~\ ; i • ,

"' .i ~

:~ .. .......,

I I

' I '. ! ; ' '' t·•r

~ ~

~.r ....

~-~

'' i i i I.

6

i • : .. '" ••

' '!

I

ci .. . . ~·

t

·~

:

100 ~--~----~-----r----,-----,-----.-----r----,

80

-.,. ... ~ 60 1&.1 a:: ;::) (/)

~ 40 a:: 0..

20

100 300

-GMC Astro 9!5/ Fontaine

Ford CL 9000/ Fruehauf

400 500 600 700

TIME (msec)

(After MacAdam (22)

Figure 5. Response time histories, trailer brake chamber pressure (13).

7

800

The air pressure in the system is converted into an actuation force on the brake by the brake chamber. Typically, the air pressure displaces a piston/diaphragm assembly through a distance called the stroke. This displacement actually produces the force necessary for brake application, which force can be ideally computed by the product of the pressure times the piston/diaphragm area. In practice; however, this ideal result is not obtained. Figure 6 presents the results of processing typical test data and plotting "effective area" as a function of stroke and pressure. This graph illustrates dramatically how significant shortfalls in design force can be experienced in mal­adjusted brakes. As the brakes are used, if they are not properly adjusted for the current state of lining wear, the chamber can run out of stroke. Also, as brakes are heated through use, arum expansion may increase the influence of marginally adjusted links so that a loss of torque capacity is realized.

The brake mechanism itself is typically one of three types:

i) S-cam, i i) Wedge, or

iii) Disc.

Disc brakes are relatively new on commercial vehicles but their frequency is increasing due to improved stability, cooler operation (less fade), longer lining life, improved contamination (water) recovery and lighter weight (11).

In order to draw reasonable conclusions regarding the expected performance of a brake system, the components described above must be analyzed as a system. A detailed analysis is beyond the scope of this paper; however, excellent developments are presented in (12,13) with a more general, less technical presentation in reference (10).

The essential facts regarding truck brake operation as it relates to the current subject are as follows:

i) Proper brake diaphragm adjustment is essential to proper system operation. This effect, varying as described earlier, is at least as significant as the load variation on the assembly (12,13).

ii) The S-cam and wedge-type brake mechanisms, although inherently self-energizing and hence potentially more appealing in normal operation, are more prone to fade, less consistent, and more susceptible to wear and adjustment

·malfunctions than their disc counterparts (10,11).

iii) Conventional (non anti-lock control logic) brake systems must be designed for best compromise between fully-loaded (high temperature rise/drum expansion) and empty conditions. This can promote systems which either tend to lock at 1 ight loaas (better heavily loaded system) or ones which are marginal under full load (better brake stability and control when empty). In any event, brake adjustment is

8

30

..... "' c .... < 100 psi w 80 0: 20 < 60 0: w 40 m ~ 20 < :::1: u IOpsi

w > ~ u 10 w ... ... w

Bendix Westinghouse Type 30 Air Camber

.4 .8 1.2 1.6 2.0 2.4

STROKE (in)

(after MacAdam (13)

Figure 6. Brake chamber effective area versus stroke (13}.

9

critical and must be accomplished on a regular basis or the system should incorporate automatic slack aojusters (13,14).

iv) The "ideal" combination of an anti-lock control logic and disc brake mechanism can exploit all the advantages of each, plus be adaptable to extremes in load conditions much more readily than non anti-lock systems.

Control Loss During Emergency or Extreme Maneuvers

There have been many reports written on how braking systems can be improved in order to achieve shorter stopping distances. Whether or not shortening truck stopping distances will increase highway safety is a question that has not been definitively answered. It does seem apparent that preventing wheels from locking during extreme braking maneuvers would prevent many losses of control.

Table 1 is an effort to define many of the emergency maneuver conditions that can lead to losses of control. Factors that initiate a control loss during braking, cornering or a combination of both are described in the second column. The possible result of these attempted maneuvers is given in the third column and possible ways of recovering from the developing control loss situation are stated in the final column. It is with reservation that this final column has been prepared. The instinct and reactions of an experienced and skilled driver cannot be put in tabular form by researchers. In many cases emergency maneuvers and reaction to the resultant dynamic conditions must take place in a time period that allows only instinctive reaction rather than adherence to appropriate learned responses.

The best judgment of the writers has been exercised in developing these "possible methods of recovery" along with the council of Mr. Robert D. Ervin, Research Sctentist at the University of Michigan Transportation Research Institute.

Many of the recovery methods are obvious. There are, however, a few that may warrant some discussion. The main results of control loss are tractor jack-knifing, trailer jack-knifing and rollover. These conditions are illustrated by Figure 7. For example, when tractor drive wheels break traction and skid laterally during an excessive cornering maneuver it would appear helpful to disengage the clutch. This reduces circumferential braking either due to drive torque or engine orag and gives the tire more cornering capacity. The time factor may make this reaction impractical in most cases.

Some thought has been gfven to the idea that activating the trai1er brakes in a condition of excessive body roll or lateral acceleration could destroy or at least reduce the cornering capacity of the trailer tires and reduce the roll moment. While this may be true it would also undoubtedly cause trailer swing. Thus one control loss mode would be traded for another. In view of this, and a growing feeling in the industry that individually activated trailer brakes (hand lever trailer brakes) are an anacronism, the writers have declined to recommend the use of trailer brakes for any situation. ·while there may be some

10

TABLE I TYPES OF CONTROL LOSS DURING EMERGENCY OR EXTREME MANEUVERS ANO POSSIBLE SOLUTIONS

Maneuver Factor Initiating Result Possible Methods '*** Control Loss of Recovery

Front tractor wheels Steering control is lost Modulate** service brakes lockup and vehicle may stay on to regain steering.

a straight path in a Straight stable condition. Line Braking Tractor drive Excessive tractor yaw may Modulate service brakes

wheels lockup occur quickly. (Tractor and steer in direction jack-knifing) of movement (i.e. if

tractor is rotating to left steer to right and vice versa.)

Trailer wheels Trailer swing may occur. Modulate service brakes lockup (Trailer jack-knifing) and if reasonable

. accelerate moOestly,

Tractor front Steering control is lost. Reduce steering and "feel" wheels skid* Reduction in cornering for steering limit of tire laterally capacity and probable cornering capacity.

drift of tractor front end outside of intended curve path.

Tractor drive Reduction in cornering and Reduce steering. steer in wheels skid* probable drift of tractor direction of movement and laterally drive wheels outside of depress clutch.

intended curve path. Ex-cessive tractor yaw may occur quickly. (Tractor jack·knlfing)

Cornering Trailer wheels Drift of trailer wheels Reduce steering. skid* laterally outside of intended curve

path. Trailer swing may occur. (Trailer jack·knifing)

Excessive body roll or lateral

Rollover Reduce steering.

acceleration

Tractor front wheels Steering control is lost. Reduce steering and lockup and/or skid* Reduction in cornering modulate service brakes. laterally capacity and probable drift

of tractor front end outside of intended curve path.

Tractor drive wheels Reduction in cornering and Reduce steering and lockup and/or skid* probable drift of tractor modulate service brakes. laterally drive wheels outside of in· Steer in direction of

Combined tended curve path. Excessive movement and depress Braking tractor yaw may occur quickly. clutch. Cornering (Tractor jack·knifing)

•

••

•••

Trailer wheels Drift of trailer wheels out· Release trailer brakes lockup and/or side of intended curve path. l"i'tfie'Y have been skid* laterally Trailer swing may occur. activated. Reduce

{Trailer jack·knifing) steering and modulate service brakes.

Excessive body roll Rollover Reduce steering. or lateral accelera-tion

Lateral skidding means the cornering capacity of the tire is saturated, the tire may start to skip laterally and the cornering force may be greatly reduced.

Modulating the service brake means successively activating and releasing the foot treadle to prevent lockup while braking effectively.

In some cases on pavements having low values of available friction the result of attempted emergency maneuvers will occur so quickly the driver will not have time to provide other than an instinctive reaction. The time necessary for the theoretically best countenmeasufes will not be available and/or effective.

11

...... N

.. -:.. ... .._ -... - ..

TRACTOR REAR WHEELS

__.-LOCKED

(after Ervin (13))

TRACTOR JACK-KNIFE TRAILER JACK-KNIFE

Figure 7. Loss of Control Responses of Tractor Semitrailers.

,.....--- -.

ROLLOVER

specific situations when they could be of value, there are many more where their use would be counter productive.

Now what does modulating service brakes and "feeling" for the steering limit mean? Modulating brakes means simply the driver getting on and off the foot treadle valve in a way to prevent !nl wheels from locking and at the same time brake effectively. "Feeling" for the limit of cornering is probably not practical. Under a few conditions of almost steady state cornering a driver may be able to sense when his tires start skidding laterally and reduce and then add steering to approach the cornering limit.

It may remain for detailed dynamic handling and stability testing of a variety of rigs to determine the practicality of many of the suggested methods of recovery.

HYDROPLANING

It has been understood in the highway engineering community that large truck tires do not hydroplane at highway speeds. There were

·several reasons why this myth developed. In the early '60's, Horne and his fellow engineers in NASA discovered and studied the phenomenon of hydroplaning as it related to aircraft tires. Because of the way aircraft tires are constructed, the shape of the contact patch remains much the same for a fairly wide variation of tire load. The NASA group found that one could predict hydroplaning speed as a simple function of tire pressure. This relationship predicted hydroplaning speed of tires with 60 to 100 psi inflation pressure well above what could be achieved by highway vehicles. Since truck tires normally required pressures in this range, it was felt that they would not be subjected to speeds high enough to hydroplane. Further work in the late '60's on automobile tires confirmed that hydroplaning speeds would be extremely high at high levels of tire pressure. These studies of automobile tires, including research by Stocker and Gallaway at Texas Transportation Institute, suggested that tire loads were an unimportant variable. Those who interpreted this work to mean that truck tires could not hydroplane did not appreciate the following. While an automobile tire for a 4000 lb vehicle may have a normal range of loads from SOO to 1200 lbs, a truck tire may be operated with loads varying from bOO to 6000 lbs. With this extremely wide load variation, the aspect ratio of a truck tire surface contact zone varies spectacularly, leading to hydroplaning conditions for a lightly-loaded, albeit normally inflated, truck tire at speeds common to highway vehicles. This footprint aspect ratio is the ratio of the surface contact zone width to length.

At the Transportation Research Board's annual meeting in January of 1984, it was suggested to Committee A2B07 (Surface Properties - Vehicle Interaction) that a Task Group be set up to look into the special problems of tractor-trailer loss of control. During the course of committee discussion, Horne disclosed that he had written a paper predicting that truck tires, in an extremely low load condition, will hydroplane at highway speeds and explained why this should occur. Horne was asked if this theory had been experimentally verified, since it was definitely contrary to "conventional wisdom". Horne responded that it

13

had not been so verified. Shortly after that meeting, Horne sent Texas Transportation Institute (TTI) a copy of his forthcoming paper, scheduled for presentation at the meeting of ASTM E-17 in April. Horne's arguments, explanations, and predictions were compelling. Intrigued by the possibility of explaining why unloaded tractor-trailers are so prone to loss of control during wet weather, engineers at TTI constructed the test trailer shown in Figure 8. The hydroplaning trough used in this testing is shown in Figure g and the trailer during testing in the trough in Figure 10. The test data are summarized in Table 2.

At this time, only four data points have been determined. The lightest load available on the test tire was 940 lbs. By imprinting the tire footprint (contact area on pavement surface) using carbon paper, it was determined that the aspect ratio (the nominal ratio of the footprint width to length) was 1.4 for tire pressure varying between 20 and 100 psi. This footprint is shown in Figure 11 at an inflation pressure of 75 psi.

By gradually increasing speed, the speed was determined, for a particular load and pressure condition, at which the tire began to spin down. That point was a reauction of tire speed of 2 mph. By increasing speed beyond that value, large values of spin down could be achieved.

TAtlLE 2 - TABULATION OF TEST CONDITIONS

Wear hydroplaning Tire Condition Pressure psi Load lbs W/i Speed mph

Truck 10.00.20 New 20 940 1.40 43

II Worn* 40 940 1.40 51 II II 75 94() 1.43 58 II II 100 940 1.41 62 II II 70 3600 0.95 Over 62** II II 100 3600 1.10 Over 62**

Water depth about 1/4 in ± 0.1 in

* Worn to approximately 2/32 in tread remaining ** 62 was the top speed achievable. No spin down was detected at

this speed.

Figure 12 shows how the four data points compare to Horne's pre­dictions. Within the range of practical truck tire pressures, 60 to 120 psi, the comparison appears quite good. Horne's prediction is about four mph low (8%) at 60 psi, correct at 75 psi and about 6 mph (10%) high at 100 psi. Since there was no replication of the data achieved, this is probably within the range of experimental variation if such factors as tire construction, tire tread depth, water depth, and pavement texture are considered.

14

)

teD-. . . • J

Figure 8. Hydroplaning test trailer and towing unit.

15

Figure 9. Hydroplaning trough. Water depth was about ~;,. inch ±0.1 inch.

16

Figure 10. Test tire after spinning down due to full dynamic hydroplaning as the test wheel is pulled down a water trough.

17

t w

.. 1

Figure 11 .

Estimated Boundary

Footprint of test tire (load 940 lbs, pressure 75 psi, aspect ratio, W/z = 1.4).

18

::1: a. :E c:i' w w a. en

120~----------------------------~

100

80

60

40

20

20 40

= 1.4 HORNE at w/~~.

60

. TTl/ 23.3(PSI) 0 ·21

R2 = 0.999

w/l .. 1.4 LOAD = 940 LBS.

80 100 120

INFLATION PRESSURE, PSI

Figure 12. Comparison of TTI data points and Horne's predictions at w/z = 1.4.

19

Horne's equation is

VEL= 7.95 (P) 0·5 1 °· 5 W/ l

( 1 )

compared to an equation based on TTl's curve fit, normalized at the test aspect ratio of 1.4

VEL = 23.3 (P) 0· 21 1.4 °· 5 W/'1.

(2)

A comparison of the curves achieved using the two equations is given by Figure 13. It must be considered highly presumptuous to base an equation of four data points. In the future, TTl engineers expect to acquire more data at lower and higher tire loads. These data should allow the formation of a more reliable predictive equation. It is conclusive, however, that Horne's theoretical predictions are reasor.ably accurate and that lightly loaded truck tires do hydroplane. The remaining discussions will be based on the hydroplaning speeds predicted by equation 1.

(!) z z <C ...I Q. 0 a: c > J: ::E :;)

::E z :E

100

- 80 J: Q. ::E 60 -c w w Q. 40 (/)

a VEL. •7.115 -vP(w/ A)" 1 (HORNE)

a VEL. • 23.3(P)0·21 ( 1.4/W/ 1 )0 "5 (TTl)

" ;/' c--- --- o-

1.5

--.. f ___ c •. o

(w/ . .0 FOOTPRINT

ASPECT RATIO

20 40 60 80 100 120

INFLATION PRESSURE (PSI)

Figure 13. Comparison of Horne's and TTI's curves.

20

Minimum Dynamic Hydroplaning Speed

Reference (1) presents compelling arguments that the minimum truck tire dynamic speed on flooded pavements depenos upon the magnitude of both the tire inflation pressure and the tire footprint aspect ratio as shown in table 3 and figure 14.

Table 3 (from ref. 1)

MINIMUM DYNAMIC HYDROPLANING SPEED ON FLOODED PAVEMENTS

AIRCRAFT TIRES

constant footprint aspect ratio over load range

Vp = 10.35 P

AUTO, BUS, AND TRUCK TIRES

equation 2

variable footprint aspect ratio over load range

Vp = 7.95 P(w/~)- 1 equation 4

where

Vp min. ayn. hydroplaning speed, mph p tire inflation pressure, psi w tire footprint width, in

1 tire footprint length, in

W/ i. tire footprint aspect ratio

21

:t: ll. ~

The effects of loaded and unloaded trailers on truck tire minimum hydroplaning speed are shown in tables 4 and 5.

The data from tables 4 and 5 were obtained from tractor-trailers that duplicated actual pressure and load conditions for two jack-knifing accidents. What the tables show is a rather small change in tire loading on the front steering axle as compared with the tire loadings on the rear drive axles of the tractor for loaded and empty tractor-trailer conditions. For example, the tire loading on the front steering axle decreases approximately 20% going from tractor-trailer 1 oaded to empty conditions while the tire loading on rear drive axles of the tractor decreases by approximately 70%. The result is a small change in truck tire minimum dynamic hydroplaning speed for the tractor front steering axle tires from loaded to empty trailer conditions and a much larger change for the tractor rear drive axle tires. More significantly, a large differential exists between tractor front axle and rear drive axle tire hydroplaning speeds; from approximately 83 to 65 mph for table 3 and from 87 to 61 mph for table 5. Equation 4 (see table 3 and figure 14) was used to calculate the minimum dynamic tire hydroplaning speeds shown in tables 4 and 5. The large decrease in tire hydroplaning speeds for the tractor drive axle tires shown in tables 4 and 5 are the result of the tire footprint aspect ratio effect given by equation 4. Figure 15 shows a typical footprint variation for the forward drive axle tire (table 5) for trailer loaded and empty conditions.

120

MINIMUM DYNAMIC HYDROPL~NING SPEED

PNEUt1ATI C TIRES- FLOODED PAY Et1ENTS

0.6

- 100 c

0.8

1.0

1.2 1.4 1.6 1.8 2.0

w w :;; 80 (!)

z z 60 ct ...1 ll. o. a: 40 c ~ :t:

20

TIRE PRESSURE, PSI

Figure 14

22

FOOTPRINT ASPECT RATIO (w/ ;)

CASE 1. -TRACTOR-TRAILER FOOTPRINT CHARAC7ERIST1CS

TRAILER EMPTY AND LOADED CONDITIONS

TRAILER LOAO PER TIRE FOOTPR !NT 11YORO-

TIRE LOCATION LOAO TIRE PLANING

CONDITION LBS. WIDTH LE ~G'iH (v /1) SPEED, MPH v,IN. 1 ,IN.

TRACTOR FRONT EMPTY 3638 6.55 7.85 .a:u 82.6

STEERING AXLE P • 98 PSI LOADED 4578 8.17 9. 1 .898 79.6

TRACTOR FORWARD EHPTY 1175 5.63 4.7 1.198 65

DRIVE AXLE P • 88 PSI LOADED 4325 7.19 8.57 .839 77.6

TRACTOR REAR EMPTY 1288 5.3 4.34 1.221 64.4

DRIVE AXLE P•88PS! LOADED 4328 7 7.88 .888 75.5

TRACTOR FORWARD EMPTY 565 5.82 4.68 1.244 63.8

AXLE P•88PSI LOADED 4888 7.5 8.1 .926 73.9

TRACTOR REAR EMPTY 1288 6 6.98 .859 76.7

AXLE P • 88 PSI LOADED 5818 7.38 lli. 2 .724 83.6

TABLE 5.

CASE 2. -TRACTOR-TRAILER FOOTPRINT CHARACTERISTICS

TRAILER EMPTY AND LOADED CONOIT!ONS

TRAILER LOAD PER TIRE FOOTPRINT HYDRO·

TIRE LOCATION LOAD TIRE PLANING CONDITION LBS. WIDTH LENGTH (v/1) SPEED, MPH

v,IN. 1, IN.

TRACTOR FRONT EMPTY 4278 6.95 8.4 .83 87.3 STEERING AXLE P•188PS! LOADED 5728 7.1 9.36 .76 91.2

TRACTOR FORWARD EHPTY 1285 6.84 4.34 1.58 63.2 DRIVE AXLE LOADED 4285 7.24 8.22 .as 84.7

P • 188 PSI

TRACTOR REAR EMPTY 1128 6.76 3.88 1.74 68.3 DRIVE AXLE P • 188 PSI LOADED 3825 7.33 7.83 1.84 78

TRAILER FORWARD EMPTY 1885 4.83 4.97 .97 88.7 AXLE

P•183PS! LOADED 4275 7.47 8 .93 82.4

TRAILER REAR EMPTY 848 5.83 4.94 1.82 . 78.7 AXLE

P • 188 PSI LOADED 3978 7.28 8.42 .86 85.7

23

TRACTOR FORWARD DRIVE AXLE TIRE FOOTPRINTS 10.00-20 TIRE SIZE: P=80 PSI

TRAILER LOADED VERTICAL LOAD 4325 LBS. FOOTPRINT WIDTH 7.19 IN. FOOTPRINT LENGTH 8.57 IN. ASPECT RATIO 0.84 TRAILER EI~PTY

VERTICAL LOAD 1175 LBS. FOOTPRINT WIDTH 5.63 IN. FOOTPRINT LENGTH 4.7 IN. ASPECT RATIO 1.2

. . · ·. ~· .. ~.

'

' . .

Figure 15. Footprint Variation.

24

Truck Tire Research

The hydroplaning analysis points out thi!t high pressure tractor­trailer tires under trailer -loaded conditions have minimum hydroplaning speeds considerably in excess of the current highway speed limit of 55 mph. Therefore, 1 oaded tractor trailers are unlikely to suffer jack­knifing accidents on water puddles or flooded pavements covering highway lanes. This analysis also points out that high pressure aircraft tires have a single minimum hydroplaning speed over the complete operating range of vertical load. It is suggested that research be conducted to see if truck tire design and construction can be modified to achieve this desirable aircraft tire hydroplaning characteristic without unduly affecting other necessary truck tire requirements such as durability, long tread life, and operating safety. If such research is successful, then the current large difference between tractor front steering axle and rear drive axle tire minimum hydroplaning speeds where the trailer is empty can be reduced in magnitude, and thus reduce the threat of a jack-knifing accident occurring for this tractor-trailer load condition.

Truck Driver Education

One important obvious way to help alleviate empty tractor-trailer jack-knifing accidents on flooded pavements is to educate the driver with regard to this highway safety problem and to determine methods and techniques which the trucking industry and truck drivers can apply to minimize accident risks.

SUPPORT FROM ACCIDENT DATA

This analysis was based on the Texas portion of 1979 through 1981 Bureau of Motor Carriers Safety (BMCS) data. The analysis was limited to those accidents that involved the ICC-authorized trucks engaged in long hauls (or over-the-road service). This is the subset of truck in the BMCS file for which past accident reporting was reasonably complete. The analysis was a 1 so restricted only to those accidents occurring on 4-or-more-lane highways in Texas.

For the accident data to be supportive of the hydroplaning theory, one would expect to see a significantly higher ratio of single-truck accidents on wet pavements to those on dry pavements for empty trucks than for loaded trucks. To ensure that this higher ratio was not an artifact of the truck exposure and/or operational characteristics (e.g. empty trucks happened to trave 1 more in wet weather than did ·1 oaded trucks!), collisions involving at least one such large truck were used as a control group for the analysis.

Analysis Method

In order to analyze the proportion of tot a 1 truck accidents that were wet-weather, single-truck accidents, a discrete-multivariate model was usea. The purpose of the modeling was to account for the significant effect of truck types and day/night so that the true effect of empty/loaded on the proportion of wet-weather accidents could be

£5

obtained. In this way, the effect due to the confounding variables would be minimized and the estimates might then be stable.

The model can be expressed as follows:

log --E--1

= w + wA + wB + wCk + AB + AC + BC + ABC - p 1 J wij wik wjk wijk ( s)

where

__£__ 1 - p

is a ratio (or logit) of wet-weather to dry-weather accidents

w is the overall mean

is the main effect of empty/loaded

is the main effect of truck type

is the main effect of day/night

is the interaction between empty/loaded and truck type

and so on.

Model Estimation

Table 6 is a contingency table of the BMCS-reported truck accidents on 4-or-more-lane highways in Texas between 197~ and 19~1, cross classified by wet or dry conditions (V1), empty or loaded truck ( V2), truck type (V3), light condition (V4), and accident type (V5). There were 5 truck types defined: single-unit ·crucks (also included tractor-only), combination trucks pulling van trailers, combination trucks pulling flatbed trailers, combination trucks pulling tankers, and combination trucks pulling other types of trailers. Light condition was a dichotomous variable: day-time or night-time. The latter incluaed dawn, dusk, dark, and artificial light conditions. Variables 1 through 4 have been reported by Chira-Chaval (1984) to be significant explanatory variables of the heavy trucks' accident types. Accident type was a dichotomous variable: single-truck accidents or multi-vehicle collisions involving at least one heavy truck.

The model estimation was carried out using FUNCAT (SAS, 1982), which is an estimation procedure based on the weighted least-square principles.

The model estimation yielded the following "best" model:

26

TABLE 6

Accidents Involving ICC - Authorized Heavy Trucks on 4-or-More-Lane Highways in Texas (1979-1981)

fCC-TiPE LIG-!T TRUC<

(V5) (V4) TiPE (V3) su

v;..~

DAY FlATilED

TA'KER

OTdER

sm>LE-TRtXX su

VflN

NIGHT FlATilED

T;..·m

OTI-ER

su

VA'i I

' ' I DAY I FlATBED

I TA'KER a:U.ISICll I

I Olh'ER

' su '

taGhT FLA'i3E:J

TP.."-KER

: ""

SOURCE: Bt•lCS 1979, 1980, 1981

27

I I

I I

&l"Tf/

Lll-lDED ( V2) E L E L E L E L E L E L E L E L E L E L E L E L E L E L E L E L

E L

E L

E L

-

I

I I I

I I

I I

I •

!

PAV'i}'ENT O::!'JITICfj (Vl)

-

'n'ET

2 2

42 76 8

12 16 6

10 4 3 3

33 8)

1 6

9 4 2 2

5 2

M 162 ,-~I

31 15 11

13 -0

1' • .o 2

27 Q] -9

21 13 11

' -7

I

I

I I I I I ' I ' I

I I

LRY

4 1 7

:0

2 22

5 20

4 6

1 2

10 81

3 29

16 1 5

43 9

'?9 :m 0' J)

1"' -0.5

t,g <-~~

22 37 2.1

5'? z:::-1 --~?

149 23 .. 45

'­'

I

I

I I I !

The goodness-of-fit for this model was 17.28 for 12 degrees of freedon' (p - value = 0.1394), indicating a good fit. Table 7 shows the parameter estimates and the standard errors.

Interpretation of Modeling Result

The estimated model as represented by equation (6) indicates that the ratios of wet-pavement to dry-pavement truck accidents were significantly influenced by load status (empty/loaded), truck type, accident type (single-truck/multi-vehicle), and the interaction between load status and accident type. However, light condition (day/night) was not a significant explanatory variable.

Table 8 shows the estimated ratios of wet-pavement to dry-pavement accidents for all combinations of the significant independent variables. Figures 16(a) and (b) are the plots of these ratios for single-truck accidents and for multi-vehicle collisions, respectively. It can be seen that the ratios of wet-to-dry accidents were consistently higher for empty than for loaded trucks regardless of the accident type or the truck type. However, this difference between empty and 1 oaded trucks was far more pronounced for single-truck accidents than for multi-vehicle collisions. This differential finding was the result of the interaction between load status and accident type.

To illustrate this interaction graphically, Figure 17 shows a plot of the means of the ratios of wet-to-dry accidents for single-truck accidents and for collisions which were weighted by appropriate accident. cell frequency. If the effect of wet pavements was not particularly pronounced for empty trucks in single-truck accidents, the two lines representing single truck accidents and collisions would be parallel as indicated by the dotted line. Figure 17 indicates that the ratio of wet accidents to dry accidents for empty trucks on 4-or-more-l ane highways in Texas was on the· average about 3 times higher than expected, when collisions involving at least on.: heavy truck were used as a control group. This immediately suggests a very strong influence of wet pavements on single-truck accidents for empty trucks that was not observed for loaded trucks.

28

Table 7

Parameter Estimates and Standara Errors

TERf~ EST H1A TE I STANDARD ERROP.

w -.4815 I .0755

w2 .4445 .0617

w3 -.1785 .1819

.3883 . 0851

-.3072 .1066

-.0545 .1267

w5 .7905 .0599

W25 .2169 .0598

29

Tab 1 e 8

Estimated Ratios of Wet to Drv Accide~ts .::..::...::....:..:;:..;::_:;..::..;:. :..:..::..:....:....::..::. - - - ~ :..=::...;...;:.=~

£y Accident Type, Load Status, and Truck Type

WET/DRY RAT! 0

TRUCK TYPE LOAD STATUS SINGLE-TRUCK COLLISIONS

SINGl.E-UNIT EMPTY 2.21 0.29

LOADED 0.59 0.19

VAN EMPTY 3.89 0.52

LOADED 1. 04 0.33

FLATBED EMPTY 1. 94 0.26

LOADED 0.52 0.16 I

I TANKER HlPTY 2.50 0.33 I I I I LOADED 0.67 0.21 ! I ' I ' I I OTHER Et'>?TY 3.07 0.41 I

LOADED 0.82 0.25

30

4.0

3.5

0 3.0 .... ~ 2.5 a: >- 2.0 a: c ...... 1.5 .... ILl 3: 1.0

0.5

EMPTY L.OAOEO EMPTY L.OAOEO EMPTY I..OADEO

su VAN FLATBED TANKER OTHER

(a) Single -Truck Accidents

4.0

3.5

0 3.0 .... ~ 2.5 a: >- 2.0 a: c ...... 1.5 .... ILl

1.0 3:

0.5

EMPTY t.OADED EMPTY L.OAOEO EMPTY L.OAOEO EMPTY l.OADEO

su VAN FLATBED TANKER OTHER

(b) Collisions

Figure 16. Estimated ratio of wet to dry truck accidents.

31

w I N

0 ·--c 0:

>-... 0 ....... -CP

==

3.0

2.0

1.0

-<_(~c.'+­~'l:

c;,~o;

...,.. _______ _ Collisions

T Overrepresent at ion of wet-weather single· truck accidents for empty trucks

0~----~------------------~---'C CP 'C c 0

...J

>--c. E w

Figure 17. Means of wet to dry ratios for single-truck accidents and collisions.

CONCLUSION

Conmercial tractor semi-trailers are a major problem to control during periods of wet weather on pavements that provide limited tire-pavement friction. The characteristics of large truck tires contribute to this problem in that they also provide lower tire-pavement friction levels when compared to automobile tires. Complicating this already severe situation is the fact that truck tires on vehicles in a 1 i ghtly 1 oaded condition can undergo full dynamic hydrop 1 ani ng when a significant layer of water covers the pavement surface. This can occur at speeds easily reached by many commercial rigs.

Commercial tractor semi-trailers possess unique braking and handling characteristics. These characteristics, in combination with the wide range of load conditions experienced by these vehicles, especially the unloaded configuration, provide situations in which there is a significant probability for one or more of the following to occur:

a. Tractor jack-knifing under combinations of emergency or extreme braking and/or cornering.

b. Trailer Jack-knifing (trailer swing) under combinations of emergency or extreme braking and/or cornering.

c. Tractor or trailer initiated rolling under conditions of emergency or extreme cornering (development of high values of lateral acceleration).

There art:! ways to avoid most of these situations by driving defensively, and possible ways to recover as critical dynamic conditions begin to develop that are described here. Reducing speed in wet weather to 50 mph should preclude hydroplaning of tires inflated to 80 psi or above.

Another complication is the problem of brake system design and maintenance. Without careful attention to brake n1aintenance including adjustment, loss of braking capacity on sorr.e wheels and uneven distribution of braking effort between wheels can contribute to unstable braking. Although some authors be 1 i eve there is a best sequence of whee 1 1 ockups in a severe braking maneuver, others remain unconv i need. Some believing that insufficient evidence currently exists to establish such a preferred sequence. The problem uf brake system design under maJor load variations has traditionally dictatt:!d compromise. If the system is designed for the fully 1 oacied condition, it will not be optimum when the rig is unloaded. This is a compelling argument for anti-lock systems. The one system all authorities seem to agree is preferable to ~ seguence of wheel lockup.

Due to the wide variation in the loading conditions of tractor semi-trailer combinations, the peculiar handling characteristics of these articulated vehicles and the manner in which the braking characteristics and the loading variations effect· those ha~dling characteristics the authors are convinced that reliable, effective anti-lock brake systems will become the norm in all such vehicles in the near future. The advent of low-cost microprocessor technology with its

33

companion sensor and actuator aevelopment, leaves little doubt of the viability of this concept. The mid-1980's is witnessing the application of this technology to high-end passenger cars in both Europe and the U.S. It's use in the vehicles discussed in this paper is more compelling for safety reasons and far more easily justified on a benefit-cost basis.

A SPECIAL NOTE TO TRUCKERS

We do not know it all. We are researchers with some, but very little, experience driving the big rigs. We are trying to understand phenomena that you may have to deal with today or tomorrow and perhaps have already dealt with yesterday. This paper is an effort to help all of us understand what can happen to your rig, why it happens and how you may be able to deal with it.

There is one thing we do know, and there is no question about it. The best protection a trucker has is his seat beTt. It keeps him in the driver's seat even after some collisions, a position where he may be able to regain control. , It keeps him conscious, even after severe collisions, jack-knifes, or rolls, in a condition to help himself get to safety. Only 6% of truckers wear their seat belts. We cannot understand why. As many as three thousand of you could save your own lives next year by taking this simple, life-saving step.

34

DEDI CATION

To Louis homer Hart, trucker, who died in the iron arm~ of ctn l& wheeler trying to avoid a jack-knifed rig on a rain slick south Texas highway, November 22, 1947.

35

REFERENCES

1. Horne, Walter B. "Predicting the Minimum Dynamic Hydroplanin:~ Speed for Aircraft, Bus, Truck and Automobile Tires Rolling on Flooded Pavements", Presentation to ASTM Committee E-17 at College Station, Texas, June 5, 1984.

2. Ivey, Don L. "Truck Tire Hydroplaning--Empirical Verification of Horne's Thesis". Presentation to the Techni ca 1 Seminar on Tire Service and Evaluation, ASTM Committee F-9, Akron, Ohio, November 1984.

3. Kummer, H. W. and V.. E. Meyer "Tentative Skia Resistance Requirements for Main Rural Highways", National Cooperative Highway ~R~es~e~a~r~c~h~P~r~og~r~a~m~R~e~pu~·r~t~3~7, Highway Research Board, 1967.

4. Ivey, Don L., Lindsay I Griffin, III, Tommy M. Newton, and Robert L. Lytton "Accident Analysis ana Prevention", Vol. 13, No. 2 pp. 83-99, June, 1981.

5. Ivey, Don L., Lindsay I. Griffin, III, et al., Research Report 221-1F, Texas Transportation Institute, November 1977.

6. Ivey, Don L., Eero K. Leht i puu and Joe W. Button, "Rainfall ana Visibility--The View from Behind the wheel", Journal of Satety Research, Vol. 7, No. 4, December 1975, pp. 156-16~.

7. Ivey, Don L., tl. M. uallaway, et Design Criteria for Minin1izing FHWA-RD-79-31, December 1979.

al., "Paven1ent and Geometric hydroplaning", Report No.

8. Morris, R. S., et al., "Field Study of Driver Visual Performance During Rainfall", Report No. UOT-HS-5-0117£:, National Highway Traffic Safety Administration, Washington, D.C., November 197b.

9. Ivey, Don L., C. J. Keese, A. h. Neill, and C. F. Brenner, Highway Research Record No. 376, 1972, pp. 40-53. Presented to the workshop on 11 Anti-:Jkid Management" at the 5Uth Annual t'1eeting ot the Highway Research Board, washington, D.C., January 19/l.

10. Fitch, James William, Motor Truck Engineering Handbook, J. v.. Fitch, Publisher, Anacortes, Washington, 1983.

11. von Glasner, £. E.., L. C. Buckman, R. DeReg1;aucourt, D. J. Knignt and william A. Hertel, "The Challenge Gf Designing and Developing Brake and Wheel Equipment for Worlowide Truck Applications", SAE Publication SP-562, Truck and Bus Meeting and Exposition, November 1982.

12. Radlinski, Richard W., S. F. Williams and John M. Machey, "The Importance of Maintaining Air Brake Adjustment", SAE Publication SP-574, Truck and Bus Meeting and Exposition, November 1982.

36

13. "Mechanics of Heavy-Duty Trucks and Truck Engineering Summer Conference Proceedings, The Michigan -College of Engineering, June 1984.

Combinations", University of

14. Gillespie, Thomas D., "Heavy Truck Ride'', SAE Publication SP-607, Society of Automotive Engineers, Warr~ndale, Pennsylvania, 1985.

15. Symons, James D., "Dynamic Sealing Systems for Commercial Vehicles", SAE Publication SP-563, Society of Automotive Engineers, Warrendale, Pennsylvania, 1984.

16. Horne, Walter B., "Predicting the Minimum Dynamic Hydroplaning Speed for Aircraft, Bus, Truck, and Automobile Tires Rolling on Flooded Pavements", Presented to ASTM Comrni ttee E-17 Meeting, Texas Transportation Institute, College Station, Texas, June 4-6, 1984.

17. Horne, Walter B., "Safety Grooving, Hydroplaning and Friction", prepared for International Grooving and Grinding Association, Suite 602, 310 Madison Avenue, New York, N.Y., 10017, May 1981.

18. Horne, Walter B., and Fredi Buhlmann, "A Method for Rating the Skid Resistance and Micro/Macrotexture Characteristics of Wet Pavements", ASTM STP 793, pp. 191-~18, February 1983.

19. Dreher, Robert C., John A. Tanner, "Experimental Investigation of the Corn~ring Characteristics of 18 x 5.5, Type VIII, Aircraft Tires With Different Tread Patterns", NASA TN D-7815, December 1974.

20. Chira-Chavala, T., "Study of Accident Experience of Large Trucks and Combination Vehicles", unpublished Ph.D. dissertation, p. 181-l30, Department of Civil Engineering, the University of Michigan, Ann Arbor, Michigan, August 1984.

21. SAS User's Guide: Basic 1982 Edition, SAS Institute, Inc., Cary, North Carolina.

22. MacAdam, C. C., et al., "A Computerized ~ioael for Simulating the Braking and Steering Dynamics of Trucks, Tractor-Semitrdilers, Doubles and Triples Combinations, User's Manual-Phase 4." Highway Safety kesearch Institute, university of l"lichigan keport No. UM-HSRI-80-58, September 1, 1980.

37