WORKING DOCUMENT

STAKEHOLDER EXPERIENCE WITH REGULATIONS 2016/427 AND 2016/646 – COMMENTS APPEAR IN OVER-TYPE MODE AND IN DEDICATED TEXT BOXES; COMMENTS EXPRESS THE VIEWS OF STAKEHOLDERS BUT MAY NOT BE REGARDED AS THE POSITION OF THE EUROPEAN COMMISSION

EN EN Verification of trip dynamic conditions and calculation of the final RDE emissions result with method 1 (Moving Averaging Window) WLTC road loads: Test WLTPhigh the possible mass is very high compared to the actual vehicle weight and trips are too soft (50% normal window cannot be complied with) adjust vehicle weight for road travel at TM WLTP or select as TM real vehicle weight: (Applies even if TMlow and TMhigh have a big difference)

Use of Test mass high (TMH) for the WLTP can lead to a huge difference between TMH and real vehicle weight, caused by the 15 % payload included in TMH. If you now measure a vehicle with a very high payload but do not have this payload in Reality on the road (when performing RDE) you will never get a valid trip, because the CO2 reference mass given by WLTP is never reached in reality.

 According WLTP, vehicles will be tested in the Predominant Mode. If no Predominant mode is available, the best and the worst case is to be measured in terms of CO2. From these two measurements the average is to be used as a basis for CO2 value declared by the manufacturer. Taking as an example a vehicle with a selectable four-wheel and without Predominant mode, it has to be measured in the 2WD and 4WD mode. From these measurements is taken as basis for the declared value, and is also basis for RDE. For vehicles with manual transmission, so far the Predominant Mode does not apply. How should this be handled looking on RDE and type approval?  Regarding WLTP type approval procedure there are different things that have to be clarified because they are effecting RDE procedure: - How to deal with cars that have a low power to mass ratio and do not fit to class 3b? Highway results for reference curve will be missing, trips will get invalid. - How to deal with cars that are type approved according the capped speed procedure? Trips will be invalid. - Matrix procedure. The non linear extrapolation during the Matrix procedure will lead to high WLTP CO2 values that will never be found in real driving. Trips will be invalid. The treatment of Multi-Stage vehicles is not defined in RDE legislative. Especially uncompleted vehicles completed after type approval by a second manufacturer cannot be tested in representative way during type approval.

The lack of transparency of what’s happening inside the evaluation tools EMROAD and CLEAR complicates the assessment of the tools and the assessment of what the corrections would be for future RDE compliant vehicle. Test results from the RDE monitoring phase, that started from 20 April 2016, have not been made available so far. Only if the test results become available to parties involved in the RDE-LDV expert group a proper assessment of the evaluation tools can be conducted.

Confused Road Load Model situation, e.g... Many places in RDE regs use Reg83 RLM: PEMS validation procedure - Appendix 3 Section 3.2.2 Power Binning de-normalisation - Appendix 6 Section 3.4.1 But other places use GTR15: Moving Averaging Windowing CO2 characteristic curve generation - Appendix 5 Section 4.1 Power Binning VeLine defnition - Appendix 6 Section 4 However where GTR15 used, there's no definition of what RLM to use - TEL / TEH / value for actual vehicle being RDE tested?

The RDE Regulation does not specify what CO2 reference and what CO2 g/km shall be used as reference calculation (Test Mass Low, Test Mass High or Test Mass Weighted).

1. INTRODUCTION The Moving Averaging Window method provides an insight on the real-driving emissions (RDE) occurring during the test at a given scale. The test is divided in sub-sections (windows) and the subsequent statistical treatment aims at identifying which windows are suitable to assess the vehicle RDE performance. The “normality” of the windows is conducted by comparing their CO2 distance-specific emissions1 with a reference curve. The test is complete when the test includes a sufficient number of normal windows, covering different speed areas (urban, rural, motorway).

Step 1. Segmentation of the data and exclusion of cold start emissions (section 4 in Appendix 4);; Step 2. Calculation of emissions by sub-sets or “windows” (sectiont 3.1); Step 3. Identification of normal windows; (section 4) Step 4. Verification of test completeness and normality (section 5); Step 5. Calculation of emissions using the normal windows (section 6).

1. SYMBOLS, PARAMETERS AND UNITS Index (i) refers to the time step Index (j) refers to the window Index (k) refers to the category (t=total, u=urban, r=rural, m=motorway) or to the CO2 characteristic curve (cc) Index “gas” refers to the regulated exhaust gas components (e.g. NOx, CO, PN)

Δ - difference ≥ - larger or equal # - number % - per cent ≤ - smaller or equal a1, b1 - coefficients of the CO2 characteristic curve a2, b2 - coefficients of the CO2 characteristic curve dj - distance covered by window j [km] - weighing factors for urban, rural and motorway shares h - distance of windows to the CO2 characteristic curve [%]

1 For hybrids, the total energy consumption shall be converted to CO2. The rules for this conversion will be introduced in a second step.

3 - distance of window j to the CO2 characteristic curve [%] - severity index for urban, rural and motorway shares and the complete trip k11, k12 - coefficients of the weighing function k21, k21 - coefficients of the weighing function MCO2,ref - reference CO2 mass [g] - mass or particle number of the exhaust component “gas” [g] or [#] - mass or particle number of the exhaust component “gas” in window j [g] or [#] - distance-specific emission for the exhaust component “gas” [g/km] or [#/km] - distance-specific emission for the exhaust component “gas” in window j [g/km] or [#/km] Nk - number of windows for urban, rural, and motorway shares P1, P2, P3 - reference points t - time [s] th t1,j - first second of the j averaging window [s] th t2,j - last second of the j averaging window [s] - total time in step i [s] ti,j - total time in step i considering window j [s] - primary tolerance for the vehicle CO2 characteristic curve [%] - secondary tolerance for the vehicle CO2 characteristic curve [%] - duration of a test [s] v - vehicle speed [km/h] - average speed of windows [km/h] - actual vehicle speed in time step i [km/h] - average vehicle speed in window j [km/h] average speed of the Low Speed phase of the WLTP cycle - - average speed of the High Speed phase of the WLTP cycle average speed of the Extra High Speed phase of the WLTP cycle - w - weighing factor for windows - weighing factor of window j

2. MOVING AVERAGING WINDOWS 1.1. Definition of averaging windows The instantaneous emissions calculated according to Appendix 4 shall be integrated using a moving averaging window method, based on the reference CO2 mass. The principle of the calculation is as follows: The mass emissions are not calculated for the complete data set, but for sub-sets of the complete data set, the length of these sub-sets being determined so as to match the CO2 mass emitted by the vehicle over the reference laboratory cycle. The moving t average calculations are conducted with a time increment corresponding to the data sampling frequency. These sub-sets used to average the emissions data are referred to as “averaging windows”. The calculation described in the present point shallmay be run from the last point (backwards) or from the first point (forward). JRC: To remove ambiguity the order of calculating windows should be specified; windows should be calculated from the first point (forward) to ensure correct inclusion of cold-start. To avoid possibly different results when running the calculation in opposite directions.

The MAW definition can be determined by two ways, backward and forward: it sometimes provides differences on the emission results and also it can be invalid in one way but valid in the other way. E.g. result on vehicle tested (petrol vehicle, not developed for RDE requirements, no type approval according to 2016/427 and after): - CO: 530mg/km forward and 393mg/km backward - NOx 97.6mg/km forward, and 91.9mg/km backward Is the choice between the two results left to the TS appreciation?

The following data shall be excluded fromnot be considered for the calculation of the CO2 mass, the emissions and the distance of the averaging windows:

The periodic verification of the instruments and/or after the zero drift verifications; – The cold start emissions, defined according to Appendix 4, point 4.4;

JRC: Does this provision open a loophole in the regulation? Instruments should function correctly for the period of a test without need for periodic verification; if needed, verification could be done at vehicle speeds <1 km/h. Are moving average windows made after exclusion of data which are calibration, cold start, vehicle ground spped < 1km/h, and engine is switched off? Method 1 for building the windows should only be allowed in one direction in order to make the procedure clearly. (Now both directions are allowed and you can take the one where the trip is valid – looking on the values afterwards will not let you know in which way the validation was made.

– Vehicle ground speed < 1 km/h with engine on or off;

ACEA recommends to delete the sentence “any section of the test during which the combustion engine is switched off” to allow advanced vehicle operation modes called “sailing” where the combustion engine is switched off during coasting.

– Any section of the test during which the combustion engine is switched off.

The mass (or particle number) emissions shall be determined by integrating the instantaneous emissions in g/s (or #/s for PN) calculated as specified in Appendix 4.

Figure 1 Vehicle speed versus time - Vehicle averaged emissions versus time, starting from the first averaging window

5 First Window

Duration of first window

Figure 2 Definition of CO2 mass based averaging windows

The duration of the jth averaging window is determined by: where:

is the CO2 mass measured between the test start and time (t2,j) [g];

is the half of the CO2 mass [g] emitted by the vehicle over the Worldwide harmonized Light vehicles Test Cycle (WLTC) deescribed in the UNECE Global Technical Regulation No. 15 - Worldwide harmonized Light vehicles Test Procedure (ECE/TRANS/180/Add.15; Type I test, including cold start);

shall be selected such as:

where Δt is the data sampling period. The CO2 masses are calculated in the windows by integrating the instantaneous emissions calculated as specified in Appendix 4 to this Annex.

2.1. Calculation of window emissions and averages The following shall be calculated for each window determined in accordance with point 3.1., - The distance-specific emissions for all the pollutants specified in this annex; - The distance-specific CO2 emissions MCO2,d,j; - The average vehicle speed

3. EVALUATION OF WINDOWS 1.1. Introduction

The reference dynamic conditions of the test vehicle are set out from the vehicle CO2 emissions versus average speed measured at type approval and referred to as “vehicle CO2 characteristic curve”. To obtain the distance-specific CO2 emissions, the vehicle shall be tested on the chassis dynamometer by applying the vehicle road load settings as determined following the procedure prescribed in Annex 4 of the UNECE Global Technical Regulation No. 15 - Worldwide harmonized Light vehicles Test Procedure (ECE/TRANS/180/Add.15). The road loads shall not account for the mass added to the vehicle during the RDE test, e.g. the co-pilot and the PEMS equipment. The Test Mass (TM) of the reference WLTC is not clearly defined. UTAC considers it is the weight of the individual vehicle with its options (and 90% filled of the fuel tank), without PEMS and without any measurement device + 15% of the pay mass load as defined in GTR15. The RDE test mass, as requested in appendix 8 is the weighing of the tested vehicle with the PEMS and power supply, the driver and the TS inspector, if he's not the driver. It is done after the test because of the cold start to take into account in the measurement, the purpose is to start the engine and leave the facility. For the future 2016/xxx EU-WLTP: 5.1.2. For the purpose of testing some artificial payload may be added as long as the total mass of the basic and artificial payload does not exceed 90 % of the sum of the “mass of the passengers” and the “pay-mass” defined in points 19 and 21 of Article 2 of Commission

7 Regulation (EU) No 1230/2012 (1). => The vehicle mass with additional mass could be higher than the TMH, and over the +3g/km tolerance of the interpolation family. PMR = rated power divided by reference mass, according to 2016/427, this is not consistent with the WLTP (rated power divided by mass in running order) nor with the 540/2014 (noise level).

3.1. CO2 characteristic curve reference points

The reference points P1, P2 and P3 required to define the curve shall be established as follows:

3.1.1. Point P1 (average speed of the Low Speed phase of the WLTP cycle) “Vehicles of categories M2 and N2: vP1 = [tbd]”

= Vehicle CO2 emissions over the Low Speed phase of the WLTP cycle x 1.2 [g/km]

3.1.2. Point P2 3.1.3. (average speed of the High Speed phase of the WLTP cycle) “Vehicles of categories M2 and N2: vP2 = [tbd]”

= Vehicle CO2 emissions over the High Speed phase of the WLTP cycle x 1.1 [g/km]

3.1.4. Point P3 3.1.5. (average speed of the Extra High Speed phase of the WLTP cycle) “Vehicles of categories M2 and N2: vP2 = [tbd]”

= Vehicle CO2 emissions over the Extra High Speed phase of the WLTP cycle x 1.05 [g/km] The calculation needs to consider the splitting of the speed requirements in M1, N1 and M2, N2 vehicles (see ANNEX IIIa, paragraph 6.3-6.5). According to that, the speed values have to be amended. Reasonable values should be evaluated by JRC or TU Graz.

Reference point in urban appears too high – urban windows for large cars show much lower CO2 emissions than the reference curve.

Because of driving WLTC in cold vehicle condition, you have a coldstart reference value for the urban part (MAW). The consequence while driving RDE is, that the MAW in urban area is too small and not normal, because of the high cold start reference value. If you have a big volume engine the problem is even bigger. Additional the problem is bigger if you have a gasoline-engine. Our proposal to solve the problem is to lower the factor 1.2 for the urban part to about 1.0 and maybe have different factors for diesel and gasoline engines (coldstart effect is much bigger at gasoline-engines). The WLTC high speed “phase 4” contains one single trip of around 8 km, during which the vehicle accelerates from 0 km/h up to more than 120 km/h and finally brakes. The motorway section of a typical RDE route is about 3 times longer and usually starts after the rural section once the vehicle is launched and running at high speed. Theoretically this makes the CO2 g/km to be higher in the WLTC compared to a stabilized speed in the RDE route. Additionally, the reference WLTC phase 4 is increased by 5%. This makes most of the motorway windows in flat or downhill sections to be below the CO2 reference curve.

These multiplication factors should be reviewed. JAMA understands that these multiplication factors itself were not clearly justified during the discussion of RDE 1st package. As a matter of fact, in many cases these multiplication factors were an obstacle to achieve a valid RDE trip despite having a normal driving behavior. In such cases unusual driving (e.g. adding unnecessary acceleration/deceleration to intentionally worsen the CO2 emissions) was needed to satisfy normality requirements. This is not in line with the intention of RDE. WLTP should be representative enough of the EU market due to the minimized flexibility of R/L setting and typical load (15% for M1, 28% for N1). Since the CO2 emissions under the WLTP and normal RDE driving is more or less the same level, the current multiplication factors seem to be too large.

3.2. CO2 characteristic curve definition

Using the reference points defined in section 4.2, the characteristic curve CO2 emissions are calculated as a function of the average speed using two linear sections ( and (). The section () is limited to 145 km/h on the vehicle speed axis. The characteristic curve is defined by equations as follows: For the section (:

For the section (:

9 Figure 3

Vehicle CO2 characteristic curve

3.3. Urban, rural and motorway windows 3.3.1. Urban windows are characterized by average vehicle ground speeds smaller than 45 km/h, “for vehicles of categories M1 and N1 resp. smaller than [tbd] km/h for vehicles of categories M2 and N2,” 3.3.2. Rural windows are characterized by average vehicle ground speeds greater than or equal to 45 km/h and smaller than 80 km/h, “for vehicles of categories M1 and N1 resp. greater than or equal to [tbd] km/h and smaller than [tbd] km/h for vehicles of categories M2 and N2,” 1.1.1. Motorway windows are characterized by average vehicle ground speeds greater than or equal to 80 km/h and smaller than 145 km/h“for vehicles of categories M1 and N1 resp. greater than or equal to [tbd] km/h and smaller than [tbd] km/h for vehicles of categories M2 and N2.”

The calculation has to consider the splitting of the speed requirements in M1, N1 and M2, N2 vehicles (see ANNEX IIIa, paragraph 6.3-6.5). According to that, the speed values have to be amended. Reasonable values should be evaluated by JRC or TU Graz. Figure 4 Vehicle CO2 characteristic curve: urban, rural and motorway driving definitions

5

4. VERIFICATION OF TRIP COMPLETENESS AND NORMALITY

1.1. Tolerances around the vehicle CO2 characteristic curve

The primary tolerance and the secondary tolerance of the vehicle CO2 characteristic curve are respectively and tol2 = 50 %. 4.1. Verification of test completeness The test shall be complete when it comprises at least 15% of urban, rural and motorway windows, out of the total number of windows. The Work Package 3 draft relaxes the minimum of 15% of motorway windows down to 5% for vehicles N2 limited to 90 km/h. This relaxation might not be enough for some routes where the motorway section starts after a stop (tolls, barriers…) and it might take long time to reach the first window at average speed above 90 km/h.

4.2. Verification of test normality The test shall be normal when at least 50% of the urban, rural and motorway windows are within the primary tolerance defined for the characteristic curve. If the specified minimum requirement of 50% is not met, the upper positive tolerance tol1 may be increased by steps of 1 percentage point until the 50% of normal windows target is reached. When using this approach, tol1 shall never exceed 30%. Some vehicles may rather require a tolerance at the lower side to meet the minimum requirement of 50%.

It is difficult sometimes to find routes with more than the 50% of normal windows and

11 in many cases the “tol1” has to be increased up to 30% in order to bring more windows of a particular section within boundaries of normality. However, we understand that these tolerances should not be increased up to 50%. The Moving Average Window methodology compares emissions to the WLTC emissions. We think that the best way to ensure that the emissions are comparable to the WLTC in terms of usage of the same engine points is by making the distance specific fuel consumption similar to what it is obtained from the WLTC.

5. CALCULATION OF EMISSIONS 1.1. Calculation of weighted distance-specific emissions The emissions shall be calculated as a weighted average of the windows distance-specific emissions separately for the urban, rural and motorway categories and the complete trip. “In the following calculations tol1 shall be 25% and tol2 shall be 50%.”

When normality requirement is achieved by increasing tol1, weight factor =1 area (±25%) should be keep to ±25% to normalize PEMS data correctly.

The weighing factor wj for each window shall be determined as such: If

Then wj = 1 If 𝑀𝐶𝑂2,d,𝐶𝐶൫𝑣𝑗ҧ൯. ሺ1 + 𝑡𝑜𝑙1/100ሻ < 𝑀𝐶𝑂2,d,𝑗≤ 𝑀𝐶𝑂2,d,𝐶𝐶൫𝑣𝑗ҧ൯. ሺ1 + 𝑡𝑜𝑙2/100ሻ

Then wj = k11hj + k12 With k11 = 1/(tol1 – tol2)

and k12 = tol2/(tol2-tol1) formulas for calculating k11 , k12 , k21 and k22 faulty k11 = 1 / (TOL1 - TOL2) k12 = TOL2 / (TOL2 - TOL1) k21 = 1 / (TOL2 - TOL1) k22 = k21 = TOL2 / (TOL2 - TOL1)

If 𝑀𝐶𝑂2,d,𝐶𝐶൫𝑣𝑗ҧ൯. ሺ1 − 𝑡𝑜𝑙2/100ሻ ≤ 𝑀𝐶𝑂2,d,𝑗< 𝑀𝐶𝑂2,d,𝐶𝐶൫𝑣𝑗ҧ൯. ሺ1 − 𝑡𝑜𝑙1/100ሻ

Then wj = k21hj + k22

with k21 = 1/(tol2 – tol1)

and k22 = k12 = tol2/(tol2-tol1)

If 𝑀𝐶𝑂2,d,j < 𝑀𝐶𝑂2,d,𝐶𝐶൫𝑣𝑗ҧ൯. ሺ1 − 𝑡𝑜𝑙2/100ሻ or 𝑀𝐶𝑂2,d,𝑗> 𝑀𝐶𝑂2,d,𝐶𝐶൫𝑣𝑗ҧ൯. ሺ1 + 𝑡𝑜𝑙2/100ሻ

Then wj = 0 where:

Figure 5 Averaging window weighing function

1

0

5.1. Calculation of severity indices The severity indices shall be calculated separately for the urban, rural and motorway categories.

and the complete trip:

13 where , are equal to 0.34, 0.33 and 0.33 respectively.

5.2. Calculation of emissions for the total trip Using the weighted distance-specific emissions calculated under point 6.1, the distance- specific emissions in [mg/km] shall be calculated for the complete trip each gaseous pollutant in the following way:

And for particle number:

Where , are respectively equal to 0.34, 0.33 and 0.33. 6. NUMERICAL EXAMPLES 1.1. Averaging window calculations Table 1 Main calculation settings [g] 610 Direction for averaging window calculation Forward Acquisition Frequency [Hz] 1

Figure 6 shows how averaging windows are defined on the basis of data recorded during an on-road test performed with PEMS. For sake of clarity, only the first 1200 seconds of the trip are shown hereafter. Seconds 0 up to 43 as well as seconds 81 to 86 are excluded due to operation under zero vehicle speed. The first averaging window starts at t1,1 = 0s and ends at second t2,1 = 524s (Table 3).

15 Figure 6 Instantaneous CO2 emissions recorded during on-road test with PEMS as a function of time. Rectangular frames indicate the duration of the jth window. Data series named “Valid=100 / Invalid=0” shows second by second data to be excluded from analysis. 160 800 j=1-43 j=100 j=200 j=559

] 140 700 h / m k [

d 120 600 ] e e w ] p o g 0 S [

d

= e n e l d 100 500 i i c c l i w n a h / e v e r g n [ e V I

f 2 / / e

O / 80 400 r

0 C ]

0 W d 1 m A e = k t / d a M i g l l [ 60 300 2

u a s O V m n C u o i C s s i 40 200 m e

2 O

C 20 100

0 0 0 200 400 600 800 1000 1200 Time [sec]

Vehicle Speed [km/h] CO2 MAW [g/km] CO2 j-th MAW [g/km] Cumulated CO2 j-th MAW [g] Valid=100 / Invalid=0 CO2 reference [g]

6.1. Evaluation of windows Table 2

Calculation settings for the CO2 characteristic curve

CO2 Low Speed WLTC x 1.2 (P1) [g/km] 154

CO2 High Speed WLTC x 1.1 (P2) [g/km] 96

CO2 Extra-High Speed WLTC x 1.05 (P3) 120 [g/km]

Reference Point

The definition of the CO2 characteristic curve is as follows: For the section (: with

a1 = (96*1.1 - 154*1.2)/(56.6 - 19.0) = -2.106 and : b1 = 154 – (-1.543)x19.0 = 154 + 29.317 = 183.317

For the section (:

with a2 = (120*1.05 – 96*1.1)/(92.3 - 56.6) = 0.571 Correction according to Section 4.2 of this Appendix.

and : b2 = 96 – 0.672x56.6 = 96 – 38.035 = 57.965

Examples of calculation for the weighing factors and the window categorisation as urban, rural or motorway are:

For window #45:

The average speed of the window is lower than 45 km/h, therefore it is an urban window. For the characteristic curve:

Verification of:

Leads to :

For window #556:

The average speed of the window is higher than 45 km/h but lower than 80 km/h, therefore it is a rural window. For the characteristic curve:

Verification of: 𝑀𝐶𝑂2,d,𝐶𝐶൫𝑣𝑗ҧ൯. ሺ1 − 𝑡𝑜𝑙2/100ሻ ≤ 𝑀𝐶𝑂2,d,𝑗< 𝑀𝐶𝑂2,d,𝐶𝐶൫𝑣𝑗ҧ൯. ሺ1 − 𝑡𝑜𝑙1/100ሻ

𝑀𝐶𝑂2,d,𝐶𝐶ሺ𝑣ҧ556 ሻ. ሺ1 − 𝑡𝑜𝑙2/100ሻ ≤ 𝑀𝐶𝑂2,d,556 < 𝑀𝐶𝑂2,d,𝐶𝐶ሺ𝑣ҧ556 ሻ. ሺ1 − 𝑡𝑜𝑙1/100ሻ

Leads to:

17 𝑎𝑛𝑑 𝑘22 = 𝑘12 = 𝑡𝑜𝑙2Τሺ𝑡𝑜𝑙2 − 𝑡𝑜𝑙1ሻ = 50/(50 − 25) = 2

Table 3 Emissions numerical data Window [#] [s] [s] [g] [g]

1 523 524 609.06 610.22 2 523 524 609.06 610.22 … … … … 43 523 524 609.06 610.22 44 523 524 609.06 610.22 45 523 524 609.06 610.22 46 524 525 609.68 610.86 47 524 525 609.17 610.34 … … … … 100 563 564 609.69 612.74 … … … … 200 686 687 608.44 610.01 … … … … 102 474 1024 609.84 610.60 5 103 475 1029 609.80 610.49 0 … … … 117 556 1173 609.96 610.59 4 117 557 1174 609.09 610.08 5 117 558 1176 609.09 610.59 7 118 559 1180 609.79 611.23 1

19 6.2. Urban, rural and motorway windows - Trip completeness In this numerical example, the trip consists of 7036 averaging windows. Table 5 lists the number of windows classified in urban, rural and motorway according to their average vehicle speed and divided in regions with respect to their distance to the CO2 characteristic curve. The trip is complete since it comprises at least 15% of urban, rural and motorway windows out of the total number of windows. In addition the trip is characterized as normal since at least 50% of the urban, rural and motorway windows are within the primary tolerances defined for the characteristic curve. Table 4. Verification of trip completeness and normality Driving Conditions Numbers Percentage of windows All Windows Urban 1909 1909/7036*100=27.1 >15 Rural 2011 2011/7036*100=28.6 >15 Motorway 3116 3116/7036*100=44.3 >15 Total 1909+2011+3116=7036 Normal Windows Urban 1514 1514/1909*100=79.3 >50 Rural 1395 1395/2011*100=69.4 >50 Motorway 2708 2708/3116*100=86.9 >50 Total 1514+1395+2708=5617