Detection distances to obstacles on the road seen through windscreens in different states of wear

I VTIratt

1988

Detection distances to obstacles

on the road seen through

wind-screens in different states of

wear

Gabriel Helmers and Sven - Olaf Lundkvist

db

Vag-06/7

Statens veg- och trafikinstitut (VTI) - 58 1 0 1 Linke'ping

l/TIra

1988

Detection distances to obstacles

on the road seen through

wind-screens in different states of

wear

Gabriel Helmers and Sven - Olaf Lundkvist

(db

Vég- 06/]

Statens véig- och trafikinstitut (VT/l - 58 7 0 1 Linkb'ping

Samhall Klintland Grafiska, Linképing 1988

CONTENTS Page ABSTRACT I SUMMARY II 1 BACKGROUND 1 2 PROBLEM 3 3 METHOD 4

3.1 Full-scale simulation and over all conditions 4 3.2 Experimental situation 4

3.3 Instructions 5

3.4 Experimental equipment 6 3.5 Sample of windscreens 6 3.6 Visual targets 7 3.7 Measurements of the straylight level 7 3.8 Ambient illumination 8 3.9 Headlight conditions 8

3.10 Subjects 9

3.11 Design 10

3.12 Experimental plan 10

4 EXPERIMENTAL CONDITIONS AND RESULTS,

EXPERIMENTS 1-4 11 4.1 Experiment 1 11 4.2 Experiment 2 15 4.3 Experiment 3 19 4.4 Experiment 4 23 5 DISCUSSION 28 REFERENCES 32 APPENDIX 1 APPENDIX 2 APPENDIX 3 APPENDIX 4

PREFACE

The research work presented in this report has been sponsored by Pilkington Sékerhetsglas AB. Mr G. Ullner with Pilkington has initiated the project and he and his consultant Dr. S. Lofving were of great help in the progress of the work.

The Air Force base F13M has kindly put one of their taxi-ways at our disposal for the full-scale experiments. Robo AB has kindly delivered the headlights used in the experiments.

Gabriel Helmers, project leader, is the main author of the report, while Sven-Olof Lundkvist had the main responsibility for the work related to analysis of data and light measurements

of windscreens.

But the work as a whole has been a team-work with a number of participants. Uno Ytterbom had a very important role, especially concerning the acquisition of the detection distance data. Christina Ruthger played another important role by editing the manuscript and by trying to make my English comprehensible. Besides those mentioned, a large number of people inside as well as outside the Institute have participated by giving me valuable

advice.

Linkoping in December, 1988

Gabriel Helmers

Detection distances to obstacles on the road seen through wind-screens in different states of wear.

by Gabriel Helmers & Sven-Olof Lundkvist

Swedish Road and Traffic Research Institute (VTI)

8 581 01 LINKCPING Sweden

ABSTRACT

The importance of windscreen wear for driver visibility in vehicle lighting has been studied. Windscreen wear is related to the straylight level of the windscreen. Light measurements of straylight levels in windscreens were carried out by a special instrument and by a laboratory method. Detection distances in vehicle lighting to targets on the road were studied in a series of full-scale experiments. In these experiments opposing situa-tions between two vehicles on the road were simulated. The wind-screens were exchanged between trials.

II

Detection distances to obstacles on the road seen through winds screens in different states of wear.

by Gabriel Helmers & Sven-Olof Lundkvist

Swedish Road and Traffic Research Institute (VTI)

8-581 01 LINKoPING Sweden

SUMMARY

This is a final report of a project where driver visibility in vehicle lighting has been studied in relation to windscreen wear. Windscreen wear is related to the straylight level of the

windscreen.

In the first part of the project windscreen wear was evaluated in the laboratory. Contrast reductions of worn windscreens were studied in relation to the reductions of a new windscreen and those of the human eye (according to the formula by Holladay and Stiles). The result was that the straylight luminance in worn windscreens was of the same size as the straylight luminance in the human eye. A decision was therefore taken to study effects of windscreen wear on visibility in night-time driving. (See Appendix 1.)

The straylight level of windscreens was measured by a special instrument and by a laboratory method. The reliability and the validity of the special instrument were investigated. The results show that the instrument has both a high reliability (freedom of random errors) and a high validity (good predictive value related to relative detection distances in vehicle light-ing through windscreens in different states of wear). (See Appendix 2.)

III

In the second and main part of the study, detection distances in vehicle lighting to targets on the road were studied in a series of full-scale experiments. In these experiments opposing situa-tions between two vehicles on the road were simulated. Wind-screens of different states of wear were exchanged between

trials.

The data from each of four experiments carried out were very consistent. The results showed a significant decrease in the detection distances to visual targets on the road as an effect of windscreen wear. This decrease in detection distance is valid not only for opposing situations with glare but also in low beam illumination without opposing headlights. Besides that, the decrease in detection distances was related to the straylight level of the windscreens by a linear relationship.

The combined influence of windscreen wear or straylight level

with other windscreen factors as outside dirt and rain, inside

film and tint and windscreen inclination remains to be studied.

1 BACKGROUND

The optical properties of windscreens are impaired in traffic. One sign of impairment is the relative amount of scattered light or straylight luminance in the windscreen. The driver often perceives a worn or dirty windscreen as a bright screen hard to look through. This is especially the case in situations with on-coming vehicles in night-time traffic or when driving towards a low sun. This unwanted property of irregular refraction of light in windscreens is in the following called the straylight level.

Twenty years ago Allen (1969) showed that windscreens are worn

in traffic. He identified damages from wiper operations, from hand cleaning and ice scraping actions and from particles hitting the windscreen e.g. small stones and sand. Of a sample of 13 studied windscreens, he judged 4 to be unsafe for night driving.

One year later Pfeiffer (1970) reported a work in which the straylight levels of one new and two worn windscreens have been measured in the laboratory. These windscreens were then used in an experiment. In this experiment the level of straylight in each windscreen was related to measurements of reaction times. The task of the subjects was to detect objects in a static, full scale simulation of an opposing situation between two vehicles at night. There was an increase in reaction times related to an increase in the levels of straylight of these windscreens. Two out of three differences of mean reaction times between the three windscreens were reported to be significant.

The first field measurements of straylight levels of windscreens were reported by Allen (1974). He had developed a photographic method, which he used to describe the condition of windscreen wear of a sample of 40 vehicles. Not only damages of the wind-screens were evaluated but also the effects of outside as well as inside dirt. Driver visibility was studied in a full scale static opposing situation (number of visible targets in front of a stationary vehicle) as well as in a dynamic glare situation

with one subject (detection distances to a target). Reductions of visibility were reported for worn and dirty windscreens.

Finally, Allen (1974) concludes that windscreen wipers need to

be improved and that there are possibilities of resurfacing worn

windscreens.

By using a driving simulator, Rompe and Engel (1984 and 1986) have shown effects of impaired visibility due to straylight levels of windcreens. The task of the subjects was to detect and identify the orientation of a visual target and give a response as fast as possible. The contrast of the target was varied in 4 levels while the glare- source was constant. They found a de-crease in the proportion of correct answers with increasing straylight levels of the windscreens but there were no effects

on reaction times.

Some years ago an instrument for daylight field measurements of straylight levels of windscreens was developed. This work has been reported by Timmermann & Gehring (1986). They also reported results from field studies of windscreen conditions in which measurements were carried out on samples of vehicles in traffic. The results showed a slow increase in the median straylight level and a rapid increase in the variation of straylight levels of windscreens with mileage. Comparisons between a German and a Swedish sample indicated that there can be large geographical variations of windscreen wear. This conclusion has been verified by a later study. (See Schneider and co-workers 1987).

In the first part of this project, measurements of the straylight levels of windscreens were made in a laboratory set up. The results have been published in a VTI-notat (TF 55-06A)

"Contrast Reductions in Windscreens", (included in appendix no 1).

The main conclusion drawn from these results was that there ought to be a significant reduction of driver visibility when driving with a worn windscreen. This conclusion was the starting point for the second and final part of the study.

2 PROBLEM

The first part of the study shows, that there can be a large

reduction of target contrasts related to the windscreen stray-light level and the intensity of opposing light. This is espe-cially the case on the road, where visual anglesbetween an obstacle on the road and the (often glaring) headlights of an opposing vehicle are small (see Appendix 1). So, if significant reductions of driver visibility can be found in night-time driv-ing situations as an effect of the straylight levels of wind-screens, these reductions are expected to be found in situations characterised by glare from opposing traffic.

The intensity of light towards a windscreen creates a light shatter proportional to the straylight level of that windscreen. Reductions of visibility related to this light shatter are therefore predicted to increase considerably with an increasing intensity of the opposing light.

The main purpose of the second part of the study was to show if

there are measurable reductions of detection distances to

obstacles on the road, which can be related to the straylight

levels of windscreens.

The questions to be answered by this study, are stated below:

- Does straylight levels of windscreens cause reductions of driver detection distances to obstacles on the road in vehicle lighting?

This question should be answered for each of the following opposing situations in which the intensity of the opposing light has been varied according to the range of opposing light inten-sities on the road: no opposing light or opposing parking lights, opposing correctly aimed low beams, opposing misaimed, glaring low beams and opposing high beams.

If the question above is answered in the affirmative, for at least one of these opposing levels of light,

What relationship is there between the straylight levels of

windscreens and the reduction of detection distances to

obstacle on the road in vehicle headlight illumination?

3 METHOD

3.1 Full-scale simulation and over-all conditions

Detection distances in vehicle lighting to visual targets on the road were measured in full-scale opposing situations between two passenger cars (Volvo 121 Combi). The tests have been carried out on a straight, level road, closed to traffic (a taxiway to a runway at the Swedish Air Force Base F13M). All weather and road conditions have been good and stable: dry road, clear air, etc.

3.2 Experimental situation

The experimental situation was arranged as follows. An opposing vehicle was stationary in the middle of one lane of a simulated,

7 m wide, two-lane road with 1.5 m shoulders. The two lanes were

separated by a non-reflectorized centre line as the only road-marking on the road.

An experimental vehicle was driven in the opposite lane towards the stationary vehicle. The tasks of the driver were to keep a constant speed and a position of the car right in the middle of the lane. A small number of visual targets or obstacles were placed just to the right of an imaginary edge line. These targets were placed along the road at fixed distances from the stationary vehicle. See Figure l.

Five persons were seated in the experimental vehicle: Three subjects, a driver and an experimental leader, who was sitting in a special seat in the rear of the car. Each subject had a personal, noiseless push-button held by hand. The only task of the subjects was to press their push-buttons as soon as they discovered a target.

Target posifions

/

a

\

e :B(

15m

Figure l. Lay-out of the experimental situation. The stationary vehicle (A) and the experimental vehicle (B).

3.3 Instructions

When seated in the experimental vehicle, the subjects were instructed to press their push-buttons immediately after detec-tion of each target. They were alsotold that the targets were always placed on the right curb (at a fixed distance from the

centre line of the road) and that one of the targets was removed

now and then. So, the subjects had a strong expectancy of where and when the targets would be visible. The subjects were also told that if they pushed the button for detection of a target that had been removed, their fees for participating in the experiment would be reduced by 25 SEK.

This instruction creates a situation where the maximum detection

distances (with a high degree of subjective response safety)

were expected to be measured for each subject participating in the experiment.

Immediately after the instruction was given, a few runs were made to acquaint the subjects with the situation. When each subject had reported to be ready for the first experimental run, the experiment started.

3.4 Experimental equipment

Impulses from the push-buttons of the subjects, impulses from the road (from the passage of each target) and impulses from the

vehicle (distance travelled) were recorded on an analog tape I

recorder. These data jointly formed the basis on which detection distances were calculated for each trial, subject and target.

The experimental vehicle was equipped with a special windscreen holder. Thanks to this holder the procedure of changing wind-screens was easy and rapid, i.e. requiring approximately 15 seconds. Each windscreen was kept in its original position on the vehicle. The inclination angle of the windscreens to the vertical plane was about 45°.

The headlights of the two vehicles were two ordinary round 7" units in accordance with the regulation ECE R 20. They were manufactured by Robert Bosch GmbH, (type number 0 301 600 107).

3.5 Sample of windscreens

A small sample of worn windscreens was collected mostly from crapped cars, all of the same make and type as that of the experimental vehicle. The wear of these windscreens is most probably caused by ordinary traffic conditions on the Swedish road network. The sample was not chosen by random.

Two windscreens in this sample, those with the highest stray-light levels, were selected for the experiments. One new wind-screen with a normal straylight level for a new windscreen was also selected. The new windscreen had never been exposed to traffic. Two other worn windscreens with intermediate levels of straylight were also selected from the sample of worn

wind-screens .

The new windscreen had a SLI-mean value of 0.36. Most new wind-screens have SLI-mean values in the range 0.1 - 0.6. (Personal communication with Dr S. Lofving, 1988). The new windscreen is obviously an average windscreen in this respect.

According to field measurements of straylight in windscreens with Timmermann s instrwment (Timmermann and Gehring 1986, Schneider and co-workers 1987) it should be easy to find wind-screens with higher straylight levels on vehicles in traffic than the straylight level of the worst windscreen used in these experiments.

All windscreens used in this study were produced from clear glass. They were all very well cleaned.

3.6 Visual targets

The size of the visual targets or obstacles on the road was a square with a side of 0.4 m. The targets were covered by woollen cloth for maximum diffuse reflection. The colours of the targets were black, dark grey and light grey. (The corresponding lumin-ance factors were 0.03, 0.07 and 0.18 respectively, measured in

a geometry 45°/0° to the normal of the surface.)

3.7 Measurements of the straylight level

The straylight levels of the windscreens have been measured by a special instrument. All measurement values presented below are

the mean Straylight Index or SLI(-mean) units, defined by the the manual of the instrument and in several papers. (See for example Timmermann & Gehring, 1986.)

This instrument was compared with alternative measurements by a

laboratory method. Contrast reductions of a target were measured

directly by the laboratory method and their relation to compara-tive values of the instrument was calculated. There was a rela-tively good agreement of results between the instrument and the laboratory set-up. The relationship between measurements of windscreens by the two physical methods and the visibility dis-tances received in the main experiment (experiment no 4) presen-ted in the next section is also shown. These measurements and results are presented in VTI-notat TF 55-10A. (See appendix 2.)

3.8 Ambient illumination

All runs were made at low levels of ambient illumination corre-sponding to night time driving outside built-up areas. There were no disturbing light sources in the visual field besides the lights of the opposing stationary vehicle.

3.9 Headlight conditions

The experimental vehicle (with three subjects) was operated with correctly aimed low beams in all conditions but one, in which high beams were used. (The range of vertical low beam.missaim-ing: i0.05°.) Correctly aimed low beams were used because it should be the most common condition on the road. Besides that, correctly aimed low beam patterns were designed to best fulfil the requirements of the driver in opposing situations.

The light intensity of the opposing vehicle was varied in several steps in order to introduce different amounts of stray-light luminances in the windscreens of the experimental car in which the subjects were seated. The basic level of this variable

was to introduce no disturbing straylight luminance at all. In this case weak parking lights were used on the stationary vehicle. The reason for choosing very weak light sources in this condition was to inform the subjects about the position of the stationary vehicle for purposes of control of their expectation.

A second basic level of this variable was to use glaring low beams on the stationary vehicle. This condition was realised by aiming the low beams of the stationary vehicle 1.5° up. In that way the windscreen of the experimental vehicle was illuminated

by the beam pattern below the low beam cut-off. The reason for

choosing this level was a desire to introduce a bad but realis-tic level of illumination of the windscreen from misaimed low beam headlights in ordinary traffic.

Besides these two main levels of opposing light, correctly aimed low beams and high beams were used in one of the experiments.

3.10 Subjects

A total of fourteen subjects parcipitated in the experiments. Two of the subjects were employees at the Institute, the other twelve were students at the University of Linkoping, who volun-teered to participate in the experiments. The age distribution was 49 and 34 years for the employees and between 22 and 26

years for the group of students.

No visual screening was made of the subjects. The only demand was, that on a direct question they reported subjectively normal vision by day and by night with or without correction. So, the group of subjects was neither selected nor chosen by random.

A few subjects participated in more than one experiment. The reason was that one common subject and a repetition of one or

two experimental conditions between experiments make comparisons

between experimental occasions (nights, weather, etc.)

10

ful as an experimental control. None of these subjects was chosen systematically.

3.11 Design

All experiments had a factorial, within-subjects design: All levels of the independent variables of each experiment were com-bined. Each subject participated in all conditions.

All conditions were repeated four times for each experiment and for each group of three subjects. The experimental conditions were rotated according to the ABBA-principle.

3.12 Experimental plan

As very little is known about the effects of windscreen stray-light levels on detection distances to targets on the road, the empirical work began with three explorative experiments. The purpose of these studies was to design a fourth experiment, which was the main experiment of the study.

ll

4 EXPERIMENTAL CONDITIONS AND RESULTS,

EXPERIMENTS 1-4

Four full-scale field experiments were carried out. The first three were of an explorative nature. The main purpose of the explorative studies was to gain knowledge about important inde-pendent variables for drivers' visual performance in connection with variations of windscreen straylight levels. The fourth experiment was the main experiment of the series. The purpose of this, the last experiment, was to quantify the effects of wind-screen straylight levels on detection of targets on the road.

Under the headings below the independent variables and their

levels are presented for each experiment, followed by the

results. The results are shown by figures and tables.

4.1 Experiment 1

Also new windscreens have straylight levels even if they are small. It is thus interesting to know if there is a measurable reduction of visibility as an effect of introducing a new, clean windscreen between the eyes of the driver and the visual target on the road as compared to visibility without any windscreen at all.

Experimental conditions: Detection distances through a new wind-screen were compared with the corresponding detection distances without any windscreen. Each experimental run was performed at a

constant speed of 20 km/h. The reason for driving at such a low

speed was to avoid effects of discomfort in those conditions

where no windscreen was used.

The opposing situation was in accordance with Figure l. The po-sitions of the targets along the road were 0 m and 150 m in front of the stationary vehicle. All visual targets were dark grey. (Luminance factor 0.07). The experimental vehicle with three subjects had lit, correctly aimed low beams.

12

Experimental variables and levels in experiment 1 were as

follows:

Windscreens (2 levels)

-A new windscreen (SLI-value 0.36)

-No windscreen (SLI-value 0.00)

Lights of the opposing vehicle (2 levels) -Parking lights

-Glaring low beams (aiming 1.5° up)

This makes a small factorial experiment with (2*2=) 4 experi-mental conditions. Each condition is repeated 4 times which makes a total of 16 runs for one group of 3 subjects.

Results: The results of experiment 1 are shown in Figures 2 and 3. The Figures show a systematic tendency that driving with a new windscreen introduces a small reduction of visibility as compared to driving (at low speeds) without any windscreen at all. The group means of the visibility distance data over targets are shown in the upper part of Table l. The relative decrease of visibility for the new windscreen condition is shown in the corresponding part of Table 2.

For the group of 3 subjects the visibility distance to the tar-gets detected through the new windscreen was reduced to 97 per cent of that of the no windscreen condition when meeting a vehicle with lit parking lights. The corresponding figure was 98 per cent when the opposing vehicle was operated with glaring low

beams. (See Table 2.)

Comments: The small decrease in visibility of the new windscreen was expected for the situation characterised by opposing low beam glare (causing a straylight luminance in the windscreen) but not for the situation with opposing parking lights (causing practically no straylight luminance). Is this result valid or is

it a random effect of a too small set of data?

13

Opposing parking lights

Visibility

t

100» * -*< NEW0.36

Target positions I, eel i ggeme; ee

Figure 2. Detection distances with low beams without any

wind-screen (NO-condition) and with a new windwind-screen

(NEW-condition). Opposing car with lit parking lights.

Visibility

_{Opposmg glaring low beams}

distance hn)

i

100" * -*< NEWo.36

-Target posHions ' 52:; 2:e(gifween Figure 3. _{Detection distances with low beams without any}

wind-screen (NO-condition) and with a new windwind-screen

(NEW-condition). _{Opposing car with glaring low beams}

(aimed 1.5° up)

VTI REPORT 33 9A Ta bl e 1. Sum ma ry ta bl e of Re sul ts fr om Exp er im en ts 1-4. Me an de te ct io n di st an ce s (m et re s) ove r sub je ct s an d ta rg et po si ti on s Vi si bi li ty di st an ce s (m ) fo r ea ch of 4 exp er im en ts Co rr ec tl y ai me d lo w be am s vs . op po si ng : Hi gh be am s Co rr ec tl y vs . Pa rk in g li gh ts ai me d Lo w be am s ai me d l. 5° up op po si ng lo w be am s hi gh be am s Vi sua l Exp er im en t Wi nd sc re en ta rg et Bl ac k Da rk Li gh t Da rk Bl ac k Da rk Li gh t Da rk no . SL I-me an gr ay gr ay gr ay gr ay gr ay gr ay l 0. 00 10 5. 0 72 .9 2 0. 36 81 .4 94 .7 11 6. 3 50 .1 67 .1 80 .6 3. 04 67 .3 85 .9 10 6. 2 39 .7 54 .4 69 .7

15

4.2 Experiment 2

The general purpose of the second exploratory experiment was to study differences in detection distances between a new and a very worn windscreen. There was also a specific purpose: to in-vestigate if there are strong effects related to the brightness

(the luminance factor) of the targets.

Experimental conditions: Detection distances through a new wind-screen were compared with the corresponding detection distances through a very worn windscreen. Each experimental run was per-formed at a constant speed of 50 km/h. The opposing situation was in accordance with Figure 1. The positions of the targets along the road were 0 m, 150 m and 300 m in front of the sta-tionary vehicle.

The following conditions in Experiment 1 were repeated: The experimental vehicle with three subjects had lit, correctly aimed low beams. The opposing lights were weak parking lights and glaring low beams.

Experimental variables and levels of experiment 2 were as

fol-lows:

Windscreens (2 levels)

-A new windscreen (SLI-value 0.36)

-A very worn windscreen (SLI-value 3.04)

Light of the opposing vehicle (2 levels) -Parking lights

-Glaring low beams (aiming 1.5° up)

Luminance factor of the visual targets (3 levels) -Black targets (0.03)

-Dark grey targets (0.07) -Light grey targets (0.18)

VTI REPORT 339A Ta bl e 2. Sum ma ry ta bl e of Re sul ts fr om Exp er im en ts 1-4. Re la ti ve de te ct io n di st an ce s ove r sub je ct s an d ta rg et po si ti on s Re la ti ve vi si bi li ty di st an ce s in re la ti on to th e be st wi nd sc re en co nd it io n (= 10 0) fo r ea ch of 4 exp er im en ts Co rr ec tl y ai me d lo w be am s vs . op po si ng : Pa rk in g li ghts Co rr ec tl y ai me d fl ow be am s Lo w be am s ai me d l. 5° up Hi gh be am s vs . op po si ng hi gh be am s Exp er imen t no . Vi sua l W i n d s c r e e n SL I-me an ta rg et Bl ac k Da rk Li gh t gr ay gr ay Da rk gr ay Bl ac k Da rk gr ay Li gh t gr aY Da rk gr ay l 10 0 97 10 0 98 10 0 10 0 10 0 83 91 91 10 0 ₇₉ 10 0 81 10 0 86 10 0 88 91 10 0 88 87 100 82 81 10 0 _{78 75} 10 0 97 95 90 10 0 96 91 86 16

17

This makes a factorial experiment with (2*2*3=) 12 experimental

conditions. All conditions were rotated (according to the ABBA

principle), and repeated 4 times for one group of 3 subjects.

Results: The results of experiment 2 are shown in Figures 4 - 6. The Figures show a general tendency of the worn windscreen to cause a considerable reduction of visibility. This tendency can be seen for all three target luminance factors (black, dark grey and light grey) as well as for the two opposing lighting con-ditions (parking lights and glaring low beams). The mean visi-bility distances (m) over subjects and target positions are

shown in the mid-section of Table l.

The relative reduction of the visibility distances for the very worn windscreen as compared to the new windscreen is presented in the corresponding part of Table 2. The relative reductions tend to be somewhat larger for opposing glaring low beams as compared to opposing parking lights, and for black and dark

targets as compared to bright targets.

Comments: The unexpected decrement of visibility for the no opposing light condition (opposing parking lights) in experiment 1 is repeated in experiment 2 for the worn compared to the new windscreen. Worn windscreens therefore seem to introduce a general decrease (independent of level of opposing light) in visual performance in night-time driving.

The decrement in visual performance for the worn windscreen is also valid for all three target luminance levels. Hansen and

Larsen (1979) have shown that a dark grey target (luminance

factor 0.07) corresponds rather well to the median luminance factor of pedestrian clothing. The dark grey target is therefore selected to be used in the following experiments.

18

Visibility

Black rnrge rs

distance (in) A >¢- -¢< P NEW 036 x...x P WORN 3.01, _ x - x GL WORN 3.01, 140" 120 't 100 ~-80" 60-* 1+0

20

f

\ Turge r posi rions /

Figure 4. Detection distances with low beams when approaching a

vehicle with parking lights (P) and glaring low beams

(GL). Windscreen conditions NEW and WORN. Black

targets.

Visibility

_{Dark gray fargefs}

distance ( n 1" H P NEW 036 X°~~x P WORN 3.01, x -o< GL NEW 036 x . x GL WORN 3.04 140 -120 100

~-60 --

" "' I."

Targef posi rions /

Figure 5. Detection distances with low beams when approaching a vehicle with parking lights (P) and glaring low beams (GL). Windscreen conditions NEW and WORN. Dark grey

targets. VTI REPORT 33 9A

19

VH MlHy

distance (on

Ligh r gray forge rs

I} x -o< P NEW 036 x...x P WORN 3.01, __ x - x GL WORN 3.01+ 140--120 ~-100 -- 80-- 60-- 40--20-# 0 150 300 DiSfonce between f vehk es (m) \ Target positions /

Figure 6. Detection distances with low beams when approaching a vehicle with parking lights (P) and glaring low beams (GL). Windscreen conditions NEW and WORN. Light grey

targets.

4.3 Experiment 3

There were two main purposes of the third exploratory experi-ment. The first purpose was to determine whether there are large differences in performance of visibility between very worn

wind-of The

screen wear when the opposing light is varied in 4 levels corre-screens similar magnitude of wear (as expressed in SLI-mean values). second purpose was to study the effect of

wind-sponding to the whole range of opposing light intensities on the road, from parking lights to high beams.

Experimental conditions: Detection distances were studied for one new and two very worn windscreens. The opposing vehicle was operated with parking lights, correctly aimed low beams, and glaring low beams while the experimental vehicle was operated with correctly aimed low beams. In a fourth lighting condition both vehicles were operated with high beams.

20

Three subjects participated in the experiment. The dark grey target (luminance factor 0.07) was used. The opposing situation was in accordance with Figure l. The positions of the targets along the road were 0 m, 140 m, 280 m and 380 m in front of the stationary vehicle.

Experimental variables and levels of experiment 3 are listed

below:

Windscreens (3 levels)

-A new windscreen (SLI-value 0.36)

-A very wornwindscreen (SLI-value 3.04) -Another very worn windscreen (SLI-value 2.50)

Lighting conditions (4 levels)

Experimental vehicle: -correctly aimed low beams Opposing vehicle: -parking lights

-correctly aimedlow beams -glaring low beams (l.5° up) Experimental vehicle: -high beams,

Opposing vehicle: -high beams

This makes a factorial experiment with (3*4=) 12 experimental conditions. All conditions were rotated (according to the ABBA principle), and repeated 4 times.

Results: The results of experiment 3 are shown in Figures 7 10. Firstly, the Figures show a considerable reduction of visibility for both the worn windscreens as compared to the new windscreen. Secondly, there are very small, if any, differences of visibili-ty between the two very worn windscreens. Mean visibilivisibili-ty dis-tances over subjects and target positions are shown in the

mid-section of Table 1.

Relative visibility reductions of the two very worn windscreens compared to the new one in the four lighting conditions are shown in the corresponding part of Table 2. There is a tendency to a larger relative decrease of visibility of the worn

21

screens with an increasing light intensity of the opposing

vehicle.

Comments: The decrement of visibility for the worn windscreen in conditions without opposing glare (opposing parking lights) in experiment 2 is repeated in this experiment for two worn as compared to a new windscreen.

There is also a further relative decrement for the worn wind-screens compared to a new windscreen as an effect of increasing intensities of opposing light from correctly aimed low beam to high beam illumination.

Visibili'fy

_{Correcily aimed low beams opposing}

disianceim)

_{parking lights}

A

* *< NBA/0.36

--0 1&5) 280 380 Dis rance between?

\Targe r posifions/ /' VEhiCles (m)

Figure 7. Detection distances with low beams when approaching a vehicle on parking lights for one new and two worn

windscreens

22

Visibility

Correctly aimed low beams opposing

d'sixnce m

correctly aimed low beams

*- - < NEW 0.36

11.0

120

--100

60

~-40

20

\\Target positions? VEh'Cles (m) Figure 8. Detection distances with low beams when approaching a

vehicle with low beams for one new and two worn

wind-screens

Visibility

Correctly aimed low beams opposing

distincem)

_{"glaring" low beams (i.5°up)}

' * -'< NEW 0.36

140

40

~\\

~ -

/

veh' l

( )

Figure 9. Detection distances with low beams when approaching a vehicle with glaring low beams for one new and two worn windscreens

23

Vigibmfy

High beams Opposing high beams

A H NEW0.36

11.0

a vehicle on high beams for one new and two worn windscreens

4.4 Experiment 4

In the preceding explorative experimentsthe effect of large differences in windscreen straylight levels on visibility has been studied. In this experiment, the main and final experiment in the series, the purpose was to show if there is a simple relation between the SLI-values of windscreens and detection distances to targets in headlight illumination.

Experimental conditions: Experiment 4 was repeated 3 times, with 3 subjects in each, which makes a total of (3*3=) 9 subjects. The reason for the increase in the number of subjects was to obtain a larger accuracy of the results.

The windscreen straylight level was varied in 4 steps by using 4

windscreens in different states of wear.

The headlight conditions were in accordance with the basic con ditions described in section 3.9 above: The experimental vehicle was operated with correctly aimed low beams and the stationary opposing vehicle with parking lights and glaring low beams.

24

The visual target was dark grey (luminance factor 0.07) The opposing situation was in accordance with Figure 1. The pbsi-tions of the targets along the road were 0 m, 140 m, 280 m and 380 m in front of the stationary vehicle.

Experimental variables and levels are as follows: Windscreens (4 levels)

-A new windscreen (SLI-value 0.36)

A worn windscreen (SLI-value 1.08)

Another worn windscreen (SLI-value 1.75)

-A very worn windscreen (SLI-value 3.04)

Lights of the opposing vehicle (2 levels) -Parking lights

-Glaring low beams (aiming l.5° up)

This makes a factorial experiment with (4*2=) 8 experimental conditions. All conditions were rotated (according to the ABBA

principle), and were repeated 4 times for each group of three

subjects.

Results: The results are presented in Figures 11 - 12 while the corresponding group means and standard deviations are reported in Appendix 3. The Figures clearly show that the curve for the detection distance of each windscreen has the same rank order as the SLI values of these windscreens. (There is an exception in just one point: The two upper curves of the opposing parking light condition have a deviating rank order for the target in the 0 m position. This is probably an effect of random errors of

measurements).

Mean visibility distances over subjects and target positions are shown in Appendix 3 and in the lower part of Table 1. The com-parative relative reductions of visibility of the three worn in relation to the new windscreen are shown in the corresponding part of Table 2.

Visib

25

ilify

Opposing parking lights

distance (m)

l

H NEW0.36

120* x -»< WORN1_75

\\\Tc1rge r positions//' VEhiClES (m)

Figure 11. Detection distances with low beams when approaching a vehicle with parking lights for one new and three worn windscreens in different states of wear.

Visibilify

Opposing "glaring." low beams

dis rnnce (m) A

* -'< NEW0.36

\\\Targe r positions//' VEhiCles (m) Figure 12. Detection distances with low beams when approaching a

vehicle with glaring low beams for one new and three worn windscreens in different states of wear

26

In Figure 13 these reductions (Relative Detection Distance, RDD)

have been plotted against the SLI-mean values for each wind-screen. The Figure shows that the RDD is related to the SLI-mean values of the windscreens by approximately linear relationships.

is one relationship for

There opposing parking lights and another for opposing glaring low beams.

The equations for the conditions characterized by opposing park-ing lights (p) and for oppospark-ing glarpark-ing low beams (g1),

respect-ively, are,

RDDp = 101.4 - 3.70*SLI RDDgl= 101.4 - 5.35*SLI is

There a significant difference (p<0.05) between the regres-sion coefficients of these equations.

RDD

(per cent)

0 opposin parkin li h rs

-' \

\

_{x opposmg glaring low beams}

90

--80

4-i

.

.

.

1

2

_{3 SLI (mean)}

Figure 13. Relative detection distances (RDD) as a function of

SLI(mean) values of windscreens for opposing parking lights and opposing glaring low beams.

27

The data of experiment 4 have also been evaluated by analyses of variance, within-subjects design. (See Keppel, chapter 16-19.)

In an over-all analysis there were significant main effects for all three variables: windscreen wear, target position and oppos-ing lights. (There were also two significant 2-factor interac-tions and one significant 3-factor interaction.) (See appendix 4.)

Two separate analyses of variance were therefore carried out, one for each condition of opposing lights. In these analyses both windscreen wear and target position were significant.

In the situation with opposing glare there is a strong inter-action between windscreen wear and positions of the targets: the difference between windscreens increases with a decrease in distance between the vehicles. (See appendix 4 and figure 12.) This is probably an effect of the increase in illumination of the windscreen when approaching a glaring vehicle. The increase in illumination causes an increase in the absolute level of straylight in the windscreens.

Comments: The results of experiments 1 - 3 are repeated in the main experiment. Firstly, there is a general decrement of visi-bility also in situations without opposing glare for worn wind-screens as compared to a new windscreen. Secondly, there is a further decrement related to intensity of opposing glare.

28

5 DISCUSSION

The results presented above show, that detection distances in vehicle lighting to obstacles on the road are generally impaired by worn windscreens. In these experiments comparisons between new and worn windscreens have shown a decrease in visibility ranging from 9 to 25 per cent for different opposing situations (see Table 2). Linear relationships have been found between the straylight level of windscreens and the decrement in detection

distance.

An important question is: To what degree could these results be generalised?

It is quite clear, that there is a general decrease in detection distances with increasing straylight levels. On the other hand, the data collection is not large enough to establish a general-ised linear relationship between the straylight level and the decrement of detection distance found in experiment 4. So, this result must be verified by further research. Neither has the true size of the effect of windscreen straylight levels on the decrement in detection distances in different opposing situa-tions been established in these experiments. The true effect can only be established by using a large and random sample drawn from the whole population of drivers as subjects. Before such an investigation has been carried out, the size of the effects received in these experiments, is a best guess.

A second relevant question to answer is, if a decrease of that size is important or not? The minimum safe stopping distance for alerted drivers, at normal speeds and in normal conditions on

rural roads (90 km/h), is about 100 m. As can be seen from Table

1 as well as from a great many night-time driving experiments over the years (see Perel and co-workers 1983) this safety criterion is seldom reached when driving with low beams. This is the case even in low beam driving without opposing glare. The

29

consequence of driving with a worn windscreen is thus, that an already unsafe driving situation becomes even worse.

Most differences in detection distances between new and worn windscreens in Table l are larger than 10 m. This fact means that the application of the brakes would start at least 10 m later in many night-time emergency situations as an effect of

windscreen wear.

The speed of a vehicle 10 m before coming to a standstill, with a maximum application of the brakes in good road surface

con-ditions, is at least 40 km/h. So, the difference between a new

and a very worn windscreen could in the worst case be a near-accident or a collision at a speed of 40 km/h.

The ultimate goal of every traffic safety effort is to reduce the number of accidents and their victims. The number of acci-dents or victims is a direct traffic safety measure while the detection distance in vehicle lighting is an indirect safety measure. It is indirect because its relation to the direct safety measure must be established. This work concerning the

effects of windscreen wear on accidents remains to be done. As a

matter of fact, the relation between direct and indirect safety measures is in most cases not sufficiently known.

A third important question is, whether the reduction of

detec-tion distances due to windscreen wear, as reported above, is

large or small when compared.to other disturbances of visibility in night-time driving.

Rumar (1974a) has reported visibility reductions as an effect of reduced light intensities due to dirt on the headlights of a vehicle. A reduction of the output from the headlights by 50 and 80 per cent reduces low beam visibility by about 10 and 25 per cent, respectively. According to Rumar (1974a), 50 and 80 per cent reductions of headlight illumination are typical values for wet and slushy road surface conditions, respectively. These

results constituted the main scientific basis for the

30

tion of headlight cleaners in Sweden. These reductions of visi-bility caused by dirt on headlights correspond well to the reductions due to windscreen wear, which are presented in this

report.

The question above can also be answered in relation to improve-ments of low beam headlights during the last decades. The intro-duction of the asymmetric low beam pattern as well as the halo-gen bulb resulted in improvements of visibility of about 5 per

cent each. (Rumar 1974b)

Therefore, the conclusion is that reductions of low beam detec-tion distances between 10 and 25 per cent are comparatively

large.

Below, some specific comments will be made concerning the results presented in this report.

Straylight levels in windscreens are measured when a strong beam of light hit the windscreen. (See appendix 1 and 2.) The stray-light level is related to the straylight luminance in the wind-screen, which can be perceived as a bright surface. When the windscreen is not separately illuminated no straylight luminance can be measured or perceived. It was therefore expected, that the reduction of visibility in the experiments should be very much related to the amount of opposing light hitting the wind-screen of the experimental vehicle with the subjects. But the data from all four experiments indicate that there is a strong effect of windscreen wear also in situations with practically no light coming from the opposing vehicle (weak parking lights). This is an unexpected result which remains to be explained.

Is there also an effect related to a decrease in visual acuity through a worn windscreen? Such an effect should be explained either directly by small errors of refraction in a worn wind-screen or indirectly by difficulties to accomodate correctly. An alternative explanation would be a reduction of the contrast between a target seen under small visual angles (a small target

31

at a long distance) or more specificly a reduction of the

con-trast gradient between target and background.

In this study, the problem has been to study the effects of

windscreen wear on visibility. Effects of dirt and rain have not been investigated, neither the effects of windscreen inclina-tion. These variables have been kept constant: Quite clean wind-screens mounted in an angle of inclination of 45° towards a vertical plane. The latter value is a small value compared to predominant inclinations in modern car design.

By combining effects of windscreen wear, outside windscreen dirt and rain, inside film and tint and large windscreen angles typical for modern cars, quite unacceptable reductions of driver visibility in traffic can be expected. Further research in this area is therefore urgent and should be made to reveal probable, serious problems of driver visibility due to windscreen factors and direct our efforts to significant improvements.

32

REFERENCES

Allen, .MQJ. (1969): Automobile Windshields - Surface

Deterior-ation. American Journal of Optometry, August, p. 594-598.

Allen, .MIJ. (1974): Windscreen Dirt and Surface Damage Effects.

Australian Road Research, Volume 5, No.6.

Hansen, E.R. and Larsen, J.S. (1979): Reflection factors for

pedestrian s clothing. Lighting Research & Technology, Vol. 11, No. 3, p. 154-157.

Keppel, G. (1982): Design & Analysis. A Researcher's Handbook. Second edition, Prentice-Hall, Inc., Chapter 16 - 19.

Perel,M., Olson,P.L., Sivak,M. and Medlin,J.W. (1983): Motor

Vehicle Forward Lighting. SAE Technical Paper Series, 830567, International Congress & Exposition, Detroit, Michigan, February 28 - March 4.

Rumar, K. (1974a): Dirty Headlights - Frequency and Visibility Effects. Ergonomics, Vol. 17, No. 4.

Rumar, K. (1974b): Visibility distances with halogen and conven-tional headlights. Scand. J. Psychol., 15, p. 21-25.

Pfeiffer, G. (1970): Der Einfluss des Streulichts in

Windschutz-scheiben auf des Wahrnehmen von Gegenstanden auf der Fahrbahn.

Zeitschrift fur Verkehrssicherheit, 16., II. Quartal, Heft 2.

Rompe, K. and Engel, G. (1984): The Influence of Scattered Light in Windshields on Driver's Vision during Night Driving. SAE Technical Paper Series, 840385, International Congress & Exposi-tion, Detroit, Michigan, February 27 - March 2.

Rompe, K. and Engel, G. (1986): Zur Wirkung von Streulicht bei klaren und getonten Windschutzscheiben auf die Sehleistung bei

Nachtfahrt. Glastech. Ber. 59, Nr.5, p. 132-138.

Schneider, 'W., Chmielarz, ML & Grauberg, A. (1987): ber die Aussagefahigkeit einer Stichprobe von 1330 Streulichtmessungen im Rahmen der Grentzwertproblematik. Factor Mensch im Verkehr, 36, Wege der Verkehrspsychologie, Festschrift zum 70. Geburtstag von Udo Undeutsch, Rot-Gelb-Gr n, Braunschweig.

Timmermann, AI and Gehring, G. (1986): Field Measurement of

Windshield Surface Wear. SAE Technical Paper Series, 861361,

Passenger Car Meeting & Exposition, Dearborn, Michigan, September 22 - 25.

Appendix

\ITI NUTAT

No: TF 55-O6A Date: 1987 08 19

Title: Contrast Reductions in Windscreens

Author: Sven-Olof Lundkvist, Gabriel Helmers, and Sten Lofving

Division: Road User and Vehicle Project no: 55316-4L

Project title: Laboratory measurements of contrast reductions of worn windscreens Sponsor: Pilkington S'eikerhetsglas AB

Distribution: free / fétt ttéd I

1 BACKGROUND

The driver's possibility of discovering an obstacle illuminated by his/her own vehicle s headlights depends among other things

on the experienced contrast between the obstacle and its back

ground. During night time traffic the dipped headlights of the

opposing vehicle introduce a dazzling that causes a sight reduc

tion because of scattering of light in the lens and chambers in

the eye. The scattering of light and thereby the dazzling is in

general larger for old than for young drivers. The dazzling

re-duces the contrast between a potential obstacle on the road and

its background which in turn reduces the possibilities of dis

covering the obstacle.

In the same way an illuminated windscreen of a car can be

thought of as a source of scattered light which also causes a

contrast reduction [1] [2]. The amount of this reduction is

expected to depend on the condition of the windscreen. A wind

screen with small grooves and craters and dirt probably causes a larger contrast reduction than a new clean one.

The purpose of this study is to compare the scattering of light

in the eye with that from car windscreens. The questions we intend to answer are:

Is the amount of scattered light from an illuminated windscreen

comparable to the light scattering in the eye itself?

Can the condition of the windscreen be of critical importance

for the driver's possibilities of discovering an obstacle?

Appendix 1

2 DEFINITIONS

The luminance contrast between a target and its background is an

important quantity when discussing the possibilities of

discovering a target. The luminance contrast is defined as [3]:

L L

where

C is the contrast between the target and its background. L0 is the luminance of the target in cd/m2

Lb is the luminance of the background in cd/m2

A dazzling source causes a so called veiling luminance LS in the eye.

LS is added to the background luminance Lb and reduces the contrast

as:

LO Lb

Lb + LS

From the formula it is evident that if the dazzling and thereby

LS is increased the contrast is reduced and the probability that

the driver discovers the obstacle is also reduced. There are

good reasons to assume that a windscreen works in a similar way

[2]. When the windscreen is illuminated by the headlights of an

opposing car at night the windscreen scatters light, which

creates a veiling luminance in the windscreen and causes a

con-trast reduction. This luminance is called the windscreen lumin

ance, L ._v

A good approximation of the contrast at the retina in this case

is:

Lo Lb

L_b + L_s + L_v

If LV<<Ls the influence from the windscreen on the contrast C is

very small and can be neglected. If on the other hand LV from a

worn windscreen is of the same order or larger than LS the wind screen causes a contribution to the contrast reduction that can

not be neglected. The windscreen should in this case cause a reduced probability for discovering an obstacle.

Appendix 1

3 EXPERIMENT SET UP

The driver's situation of illumination at night time traffic

without stationary illumination but with an opposing car has

been simulated in a laboratory set up. The headlights of the own

and opposing cars are simulated by projectors and the driver's

eyes have been simulated by a Pritchard photometer. A target consisting of a screen with a dark and a light field is located

10 m in front of the photometer. A windscreen was located bet

ween the photometer and the target.

The experiment set up is shown in Fig. 1.

Simulated own headugnt

Observer/ Q Visual object

photometer

\ \ $~

(obstacle)

Fig. 1 The experiment set up

Note that no direct light reached the photometer. The purpose is

to eliminate measurement errors caused by straylight in the

photometer.

4

The reduction in contrast between the dark and the light fields

on the target was used to calculate the veiling luminance in the windscreens. The independent variables were:

The condition of the windscreen: O: optimal, no windscreen 1: new clean windscreen 2: worn washed windscreen 3: very worn washed windscreen

The angle between the dazzling light source and the direction of observation, v:

1°, 2°, 3°, 4°, 5°, 10°, and 15°.

This means that the total number of combinations of angles and

windscreens were 4*7=28.

As mentioned earlier, the purpose of this study was to compare the contrast reduction at dazzling caused by the windscreen with

that caused by the eye itself. The veiling luminance caused by

the eye cannot be measured directly, but can be calculated by

formulas derived from a great number of empirical data. We have chosen the mostly used formula, namely that of Holladay and Stiles [4]:

_ 10*E s 2

where: V

S is the veiling luminance in cd/m2

is the illumination from the dazzling source in lux

v is the dazzling angle i.e. angle between the direc

tion of observation and the dazzling source in

degrees.

This formula is approximately valid for a normal young eye. The illumination at the windscreen E was also measured in order

to calculate LS.

VTI REPORT 339A

Appendix

4 RESULTS

The fields

windscreens. These measured values are presented in Fig. 2.

for different values of the angle v and for different

basic measurements are the luminances of the light and dark

m

(2)wornv ndscreen (3)very worn windscreen

100*- -~ -. __

(0) optimal (=no) (1) new windscreen

windscreen x_x__x_____x____x *x .x_____x____x

10-»

--

a»

-+

1'2' 1.

10'

15° 1'2' 4°

10°

15' 1°2 4°

10°

15' 1' ' 4°

10'

15' angle

Fig. 2 Luminance of the light and dark fields of the target

as measured through four windscreens at different

dazzling angles v. Lo=dashed line, Lb=solid line.

The measured contrast between the dark and light field of the

target for seven different dazzling angles, v and for four

different windscreens are given in Table 1. It is assumed that the dark field is the background since the normal case in vehicle illumination is that the luminance from the target is larger than that of the background (normally the road).

VTI REPORT 339A

Appendix 1

6

Since the veiling luminance of windscreen number 0 (no wind

screen) LV(O) is zero it is possible to calculate LV for the

others by means of the measured values on the contrast C. If we

define the contrast of windscreen j as C(j), the veiling lumin

ance of windscreen j as Lv(j), and the luminance of the light

and dark fields of the target as Lo(j) and Lb(j) we get:

O)

L0(0)

Lb(0)

(1)

C(

=

. LO(O) - Lb(0)

C ) = Lb<0> + chj>

(2)

Note that L0(O) and Lb(0) in equations (1) and (2) are measured

without a windscreen. C(l) is for instance lower than C(O)

because Lv(0) is zero and LV(1) is larger than zero. Assuming

that LV(O)=O and that L0(O) and Lb(0) are known we obtain:

. L0(O) Lb(0)

LV(J) = C(j) - Lb(0) (3)

Thus the windscreen luminances for the different windscreens can be calculated from the measurements of the contrast. The results are shown in table 1.

It is also possible to calculate the corresponding veiling

luminances caused by light scattering in the eye, LS for

different dazzling angles v by means of the measured value of the dazzling luminance E and Holladay s and Stiles formula.

These results are also shown in Table 1.

Appendix 1

7

Table 1 Contrast, veiling luminance of four windscreens, and

of the eye as a function of dazzling angle

Dazzling Contrast C through Veiling luminance (LV) Veiling lumin

angle v windscreen of windscreen ance of the

No. No. eye (Ls)

0 1 2 0 1 2 3 1° 13.9 2.48 0.16 0.04 0.00 7.61 168 583 1015 2° 15.8 3.74 1.00 0.17 0.00 4.59 25.8 132 254 3° 16.4 5.97 1.71 0.37 0.00 2.37 14.6 59.4 113 4° 16.4 7.59 2.94 0.64 0.00 1.57 7.67 33.3 63.4 5° 16.4 9.23 4.08 0.89 0.00 1.05 5.05 23.7 40.6 10° 16.4 12.9 9.63 4.02 0.00 0.37 1.21 4.17 10.2 15° 16.5 14.3 12.8 7.05 0.00 0.19 0.52 1.79 4.51

The contrast through windscreen 0 (no windscreen) should be

con-stant.

mainly

The fact that it is not, depends on measurement errors,

caused by dust particles in the air which scatter light

at small dazzling angles (see also Fig. 2 lower curve).

The contrast between the light and dark fields is shown as a

function of the dazzling angle v in Fig. 3.

The

reliability

the same region of the windscreen has been checked in a separate

It measurement Primarily, used has series. this

was found to be as high as 0.9998.

shows that the rather simple experiment set up fulfilled the requirements needed for calibration and

angle setting.

In another measurement series the contrast was measured at 5°

dazzling angle in 5 regions of the windscreens 2 and 3. This was

done

trast

The in

when

order to find out if there are variations in the con

measuring different regions of the same windscreen.

were calculated and the results were as follows: windscreen 2 confidence interval=4.11

windscreen 3 confidence interval=0.91

VTI REPORT 339A

i 1.19

0.08

8 Appendix 1

C=Lo LB

4x

/(0) optimal (=no)

10

\(2) worn windscreen

1' 2' 4°

10

15'

9

Fig. 3 Contrast between the dark and the white fields for

five different dazzling angles as measured with the

photometer for four windscreens

The measurements are all made in the region where the wipers

have worn the windscreen. As can be seen the variation between different measured regions is rather large for windscreen 2

while it was considerably less for windscreen 3. This shows that

the variation can be large and that it is necessary to measure

in a number of regions in order to get an accurate judgment of a windscreen.

Appendix 1

5 A FIELD STUDY

Measurements on 245 cars have been made by Pilkington Sékerhets-glas AB in order to estimate how windscreens on Swedish cars are

distributed according to scattering of light. Twenty two of

these 245 cars were excluded because their original windscreens

had been replaced.

The instrument used for these measurements is described by

Timmermann [5], and Timmermann & Gehring [6].

The measurements were made in connection with the annual inspec tion test of vehicles older than 1 year conducted by the Swedish

Motor Vehicle Inspection Company. The sample of cars chosen for

the measurements should be representative of the population of

cars from the area of the inspection station since almost no

driver refused to participate and since all cars were chosen as they arrived.

The SLI_mean and SLI_max values as defined by Timmermann were

measured [5]. The distribution of the SLImean values is shown in Figure 4.

i

Fig. 4 Distribution of SLImean values of 223 Swedish cars

measured in connectlon with the annual inspection

Appendix 1 10

Comparative measurements were made of windscreen 3 in the

laboratory set up. The SLImean value was 4.5. Figure 4 shows

that about 4% of the tested cars had windscreens of higher

SLImean values.

Since the tested sample of cars is small and does not represent

cars from other parts of Sweden we can regard this figure as a

best guess of the condition of windscreens in the population of

Swedish vehicles.

6 CONCLUSIONS

These introductory measurements on a limited number of wind

screens have shown that the condition of the windscreen is of

importance for the driver's possibility of discovering obstacles in headlight illumination especially on roads without road

lighting. The field measurements (reported in section 5) have

shown that the worst windshield chosen for the laboratory

measurements can easily be found on cars in traffic.

It was shown that the veiling luminance caused by a worn wind screen is of the same order as that caused by the eye and that

the two components roughly vary in the same way with the dazzl

ing angle.

The measured contrast reduction can on the other hand not be

evaluated in absolute values since the dazzling illuminance has

not been set at realistic levels. On the other hand relative

comparisons of contrast reductions can be made between

wind-screens as well as between windwind-screens and the human eye.

Figure 2 shows that if the dazzling source is about 1° from the

direction of observation (typical of opposing situations on the

road) the luminance level is considerably increased for both the

light and the dark fields and the contrast is thus considerably

reduced. For larger incidence angles (glare sources beside the

road) this phenomenon is reduced to a large extent.

11 Appendix 1

There are good reasons to assume that the contrast reduction

that occurs because of worn windscreens in realistic circum

stances will influence the driver s possibilities of discovering

obstacles at night time traffic. The size of this expected red

uction of driver visibility in opposing situations between cars

at night will be investigated. This will be done by full scale field experiments where the detection distances to obstacles on the road in headlight illumination are measured for windscreens

in good as well as bad conditions.

REFERENCES

[1] Padmos, P.: "Glare and tunnel entrance lighting: Effects of

straylight from eye, atmosphere and windscreen". CIE-Journal Vol.

3, No. 1, 1984.

[2] Allen, H.J.: "Windscreen dirt and Surface Damage Effects".

Australian Road Research, Volume 5, No. 6, December, 1974.

[3] Stevens, S.S.: "Handbook of experimental psychology", John Wiley & Sons, Inc., third printing, page 957, New York 1951.

[4] Vos, J.J.: "Disability glare A state of art report."

CIE-Journal, Vol. 3, No. 2, 1984.

[5] Timmermann, A.: "Direct measurements of windscreen surface wear

and the consequences for road. safety", Conference: Vision in

vehicles, Nottingham, UK, Sept. 1985.

[6] Timmermann, A. & Gehring, G.: "Field Measurement of Windshield

Surface Wear". SAE Technical Paper Series 861361, 1986.

Appendix 2

VTI-NOTAT

No: TF 55-10A Date: 1988-10-25

Title: CONTROL OF VALIDITY AND RELIABILITY OF STRAY-LIGHT LUMINANCE

MEASUREMENTS IN WINDSCREENS

Authors: Sven-Olof Lundkvist, Uno Ytterbom & Gabriel Helmers

Division: Road User and Vehicle Project no: 553 19-8

Project title: Windscreens, stage I and II Sponsor: Pilkington

Distribution: free/restréeeed/

Swedish Road and Traffic Research Institute VTI REPORT 339A

Appendix 2

Road User and Vehicle Division 1988 10-25

S-O Lundkvist U Ytterbom G Helmers

CONTROL OF VALIDITY AND RELIABILITY OF STRAY LIGHT LUHINANCE

MEASUREMENTS IN VINDSCREENS

Background

Reliability and validity are two main concepts related to each

method of measurement. These concepts can be defined as follows: Reliability: The reliability of a measurement method is the

pro-duct moment correlation (rxy) between pairs of independent

mea-surements of the same stable objects carried out by the same

method. The reliability is high when the correlation between

pairs of independent measurement values is high. Reliability is

therefore a measure of precision or freedom of random errors.

The reliability is designed p below.

Validity: The validity of a measurement method is high if it

measures that which is intended to be measured.

For example, there is a method for the measurement of stray

light in windscreens. If repeated measurements of the same

wind-screen give identical measurement values, the reliability is at

a maximum (rxy=1.00). However, we are not primarily interested

in stray light but in reductions of driver visibility caused by worn windscreens. The validity of the method is in this situa

tion expressed by the correlation between the stray light values of a windscreens and the reduction of visibility for drivers looking through these windscreens.

If the method of measuring stray light of windscreens produces values, which well predict the reduction of driver visibility