Shift Schedule Model for

Motorcycle Emissions & Fuel Economy Testing

1

Modeling Experience

• Worked on Harvard Fire Code incorporating specific subroutines to determine heat exchange through submarine walls for Electric Boat Corp

• Researched and wrote EPA’s Nonroad Emissions Model • Modeled effects on RVP of the mixing of various formulations

of oxygenated gasoline • Created a complex heat exchanger predictor model for vehicle

exhaust • Created a statistically driven motorcycle shift model

2

Background

• Presently, motorcycle manufactures are allowed to prescribe the shift schedule upon which they will operate their certificate text in the form of shift speeds 40 CFR 86-528-78

• In lieu of that, the tests can utilize the EPA’s proscribed shift speed schedule

• The development of a model based on actual rider performance would provide for a more realistic representation of shift points

9/20/2019 3

Procedure

• Motorcycle Road Testing

• Statistical Formulation

• Model Development

• Model Performance

4

Motorcycle Road Testing Route Statistics

• 1.1 hour average total time • 30.3 miles total distance

– 1.8 miles of highway (mostly on and off ramp) • 29 mph average speed • 16-57 stops depending on traffic congestion

and traffic light timing – 16 Stop Signs – 41 Traffic Lights

• 100 up-shift gear changes (average)

5

40-49 12%

AGE GROUP (YRS) 60+

8%

30-39 25%

PREFERRED RIDING INTEREST

Sport Touring 30%

Road Touring 39%

Off Road

Dual Sport 14%

20-29 29%

RIDING EXPERIENCE (YRS) 40-49

30-39 4% Less than 10

Motorcycle Road Testing

Driver Statistics – 26 Riders

6

Parameters Collected

• Date • Time • Vehicle Speed (MPH) & (KPH) • Engine Speed (rpm) • GPS Latitude and Longitude • Acceleration (mG) • GPS Altitude • Up Down Launch • Leaving and Entering Gear (Gear #) • Clutch Switch

7

Model Operation

• Three types of motorcycles are represented – Cruiser – Sport – Standard

• Different sets of sample shift patterns were developed for each type of motorcycle

• Log Normal Curves were then derived from each of these sample shift patterns

• Different Log Normal Curves were determined for each shift and for each type of motorcycle (10 curves for each type) – Up-shift 1-2, 2-3, 3-4, 4-5, 5-6 – Downshift 6-5, 5-4, 4-3, 3-2, 2-1 8

•

- - I I I I I I I I I I I I I I • ' I I • • I • • I

Sample Shift Pattern

0

20

40

60

80

100

120

140

160

180

Freq

uenc

y

Vehicle Speed Range (mph)

All Bikes – Leaving 1st Gear Upshift

Total

9

Log-N,ormal D i·stri buti ,on for 1 st lo 2nd 1Gea r llJ psh ifts o_oo .---------r-------r-------r--------.,--------r-------r-------.

0 _05

0 _04

~ =.o ~ 0 _03 e a..

0 _02

0 _01

0 110 20 30

V . hEcl 40

Sp e d (MPH)i 50 60 70

Example Log Normal Distribution

10

Model Operation

• The representative test cycle is selected (FTP, etc.) • Log-Normal curves are converted into a cumulative

distribution function (Page 12) • Model steps through each speed point incrementally and

consults the distribution function to determine if a shift will take place – As expected, the probability of a shift will increase with each

unsuccessful shift increment – Shifts will occur both up or down based on whether the test cycle

indicates that the motorcycle is speeding up or slowing down

11

0.1

0

Illustration of Typical Cumulative Distribution Function

Increasing Engine RPM

Prob

abilit

y

12

Model Options

• Operation of the model using the full probability can result in very different shift patterns for the same test procedure

• While these are viable options efforts were made to mitigate the broad range probability effects while retaining the flexibility of the model – The ability to limit the rage of the distribution was added – This allows the user to designate a percentage of the distribution range

centered around the mean – 100% (full range) to 1% (narrowest range – no variation) – Page 14 shows an example

13

,Q utp. ut V,a ri aUon in D i:sb"i buUon Ranges 0.12 r------,-----,--------,-----.------=-~------,

0.1

0.0B

~ :0 ro -...o 0.06 e

D..

0.04

0.02

Fll.llU Logoot1ma~ Djstf~lbutiolil .. 0% D1istr1ibutio.n Ra1nge. 90% D1istr1ibutio.n Ra1nge

o L-----.c:L...--~------l----JL_ ___ -1l,_ __ ____:===:c==--_J

0 10 20 JO 40 50 00

V,ehitcle S pe1ed (Ml PH)i

Distribution Range Example

14

Shifting of the Mean in the Output Distribution Range

• In order to utilize the ability to limit the distribution range and to further refine the model to allow representation of: – Aggressive Riders – mean shift to higher RPM – Average Riders – mean remains the same – Conservative Riders – mean shifted to lower RPM

• In order to perform this the “effective mean is shifted and the distribution is narrowed accordingly

15

,Q utput V.a ri .ati,on in D i:s tri buti,on R.a nges 0.12 r-----,----------r------,-----.-------=-----r----~

0.08

;e-:, :.0 ~ 0.06 Q

D..

0.04

0.02

--- Fu I Log100:rimal Distrilbut1io rn .50% rns·tr1ilbL.11tion Rairnge 90% Distr1ilbL.11tion Rainge

Q '----~-=:w.-____::_ _ ____!.L.._ __ ___!___lL ___ ...L_lL_ ___ ::::r:====::,,,-___J

0 10 20 30 40 50 60

V,ehi1cl . Sp ,edl (Ml PH )i

Representation of Mean Shift in the Distribution Range

16

Model Input

Type of Motorcycle Sport, Cruiser, Standard Idle RPM 1200

Red Line RPM 12000 Primary Drive Ratio 1.6

Final Drive Ratio 2.4 1st Gear Ratio 2.55 2nd Gear Ratio 1.9 3rd Gear Ratio 1.45 4th Gear Ratio 1.2 5th Gear Ratio 1.1 6th Gear Ratio .98

Tire Size 180/50ZR17M

Input

17

5

--

1.

., .L

UDDS eve f'R BABI LITV DI TRIBUTI l N_. . 1 r EAN

peed [M f?H) Gea

1

conds

12

7

5

3

2

1

..... ra l]J

Sample Output Conservative Rider

18

51 -_ PR UDDS BABI Lil'( DI TRIBUTI N., .5 MEAN

7

5 peed (MPH) Gea-

--...c 5 a. E -

""Cl 11.1 IJ.I c.. tn 2 3

2

1 .2 12

~onds

Sample Output Average Rider

19

5

--JC

1

2

5, ., p R El El I LI TY D I u D D s C C L TR I BU Tl N , 1 . ME N

1

E ONDS

12 1

Sample Output Aggressive Rider

20

• Both EPA (FTP) and WMTC perform upshifts and downshift at the same speed

• A Real Rider (RWDC) will not upshift and down shift at the same speed– Creates a more realistic

scenario– Model operates in this

manner

Drive Cycle GearShift Speed (km/h)

FTP WMTC RWDCAcc/Dec Cruise

Motorcycle A

12 30 23.1 11.6 2423 45 42 23.1 4434 60 53.3 42.0 6745 75 64.6 53.3 8256 N/A 75.7 64.6 10765 N/A 64.6 8354 75 53.3 5243 60 42 4432 45 23.1 2821 N/A 11.6 15

Declutch 15 10 10

Motorcycle B

Gear FTP WMTC RWDCAcc/Dec Cruise12 30 29.7 16.9 3123 45 50.4 29.7 4534 60 68.6 50.4 5845 75 85.4 68.6 7156 N/A 99.6 85.4 8165 N/A 85.4 7354 75 68.6 5843 60 50.4 4632 45 29.7 3021 N/A 16.9 19

Declutch 15 12.4 10

Motorcycle C

Gear FTP WMTC RWDCAcc/Dec Cruise12 30 30.3 22.5 3123 45 54.1 30.3 5234 60 69.9 54.1 6945 75 87.4 69.9 8156 N/A 98.6 87.4 11165 N/A 87.4 7654 75 69.9 5943 60 54.1 4932 45 30.3 3721 N/A 22.5 23

Declutch 15 15.9 10

Real World Shifting Comparison

Model - Important Distinctions

• Uses three distinct motorcycle design options which will provide different results

• Will not require a given upshift or downshift at the same speed during a given test

• Will not require downshift at the same speed as its analogous upshift

• Model could be utilized to make comparisons between manufacturer shift schedules and various riding styles

22