Biorid II Qr

8/11/2019 Biorid II Qr

1/87

MADYMO Quality Report Release Update

BioRID-II facet Q model version 3.3.1 (R7.4.1)BioRID-II fem Q model version 2.3.1 (R7.4.1)

REPORT NUMBER: QBioRIDII-120531DATE: May 2012


2/87

MADYMO Quality Report QBioRIDII-120531


3/87


c Copyright 2012 by TASSAll rights reserved.

MADYMO R has been developed at TASS BV.

This document contains proprietary and condential information of TASS. The contents of this doc-ument may not be disclosed to third parties, copied or duplicated in any form, in whole or in part,without prior written permission of TASS.

The terms and conditions governing the license of MADYMOR

software consist solely of those set forthin the written contracts between TASS or TASS authorised third parties and its customers. The softwaremay only be used or copied in accordance with the terms of these contracts.


4/87



5/87


Table of contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.1 Tests overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.2 Descriptions of the tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2.1 Head Component Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2.2 Spine Component Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2.3 Pelvis Component Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.2.4 Full Dummy Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.2.5 Rigid Seat Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3 Rating of the validation set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173.1 Overall test results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

4 Comparison of results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214.1 Range plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214.2 CPU Time comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

A Rating Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27A.1.1 Rating a MADYMO dummy model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27A.1.2 Comparing two scalar values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28A.1.3 Comparing two signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28A.1.4 Adding scores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

A.2 Additional information on calculation of dummy rating scores . . . . . . . . . . . . . . 29A.2.1 Comparing two scalar values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29A.2.2 Adding scores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30A.2.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30A.3 Results for more complex examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

B Signal results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35B.1 Signals of Head tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35B.2 Signals of Spine tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35B.3 Signals of Pelvis tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54B.4 Signals of Full tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

B.5 Signals of Rigid seat tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

i


6/87


ii


7/87

MADYMO Quality Report QBioRIDII-120531 Introduction

1 Introduction

In 2005, TASS developed new MADYMO Dummy Modules, referred to as "Quality Dummy" modules.Each "Quality" Dummy Module consists in principle of one dummy model and one quality report. Theaim of the quality report is to provide simulation engineers with the information required to judge the

predictive capability of a MADYMO occupant simulation, by objectively detailing the strengths andweaknesses of a dummy model. Different dummy model versions can be compared in the quality re-port. Comparison of different dummy models directly in one report is also possible, provided the exper-imental validation data set is identical for all dummies. The predictive power of an occupant simulationis in general determined by the quality of the occupant model(s), the quality of the environment modeland the ability of the simulation to capture the interaction between the environment and the occupant(s)accurately.

This document provides the information necessary to judge the quality of the following models:

MADYMO BioRID-II facet Q dummy model, version 3.3.1 (R7.4.1)MADYMO BioRID-II fem Q dummy model, version 2.3.1 (R7.4.1)The response of the models is compared to experimental data.

Chapter 2 of this document describes the experimental data set used to validate the dummy model,which is a sub-set of the data set used for the development. The quality of a dummy model dependshighly on the quality and completeness of its experimental validation data.

Chapter 3 gives an objective quality assessment (or rating) of the model for the data set described previ-ously. The rating denes percentage values for various aspects of "model quality", and gives an overallvalue, allowing clear oversight of quality. Rating scores are given for the complete model, subcompo-

nents, and for each experiment. Each experimental score is the sum of results of mathematical functionsthat determine the similarity of the simulation results to experimental results, taking into account thepeak value, peak timing and overall signal shape.

Chapter 4 compares simulation peak values against experimental peak values. The plots give an insightinto peak value correlation levels and the range of validation. Information on simulation runtimes canalso be found in this chapter.

A description of the rating method used in this document is given in Appendix A. This appendix givesa detailed explanation of the methods used to calculate rating scores.Appendix B shows plots comparing experimental and numerical time history signals belonging to the

validation data set described in Chapter 2. The calculated rating values are plotted in the gures as well,allowing the reader to link the mathematical rating values to the correlation visually presented in thecurves.

This quality report is released to customers purchasing a MADYMO Quality Dummy Module, and issubject to the terms of a non-disclosure agreement.

As a service, TASS can also deliver quality reports that use a validation data set that is specied andmade available by the customer. This allows the Quality Dummy customer to evaluate the quality of the dummy model using data that is representative of the type of applications typically used by the

customer in their development work. TASS will of course commit to keeping such a data set and relatedresults strictly condential.

1


8/87

Introduction MADYMO Quality Report QBioRIDII-120531

1.1 Conventions

In this report, the same colour is used for a particular model in every graph. In the time history plotsthe experimental signals are showed in black. Each model is referred to by a label. In the rangeplots adistinction is made between the different types of experiments through the use of symbols, see the table below. All rating values are expressed in percentages. The SI system is used for all signal units, exceptfor the time which is given in milliseconds (ms) instead of seconds (s), as is done in the MADYMOsolver.

Table 1.1 symbols used in this repor t

Model BioRID-II facet Q model version 3.3.1 (R7.4.1) signalcomponentfullmax peak scoremin peak scoremax peaktime scoremin peaktime scorewifac score total signal score

M2 BioRID-II fem Q model version 2.3.1 (R7.4.1) signalcomponentfullmax peak scoremin peak scoremax peaktime scoremin peaktime scorewifac score total signal score

2


9/87

MADYMO Quality Report QBioRIDII-120531 Experiments

2 Experiments

For the validation and rating of the model, many experiments have been used. This chapter presentsdescriptions of the experiments that are used for the quality rating. All information that is publiclyavailable about the tests can be found in this chapter.

In the rst section, all experiments used for the rating are listed in tables. These tables contain 8 columns.Below you can nd the description of the column headers:ID = unique testnumber#F = number of loading signals (forces and moments/torques) measured#P = number of positional signals (displacements and rotations) measured#V = number of velocities measured#A = number of acceleration signals measured#I = number of injury values rated

In the second section of this chapter, more detailed descriptions are presented in order to give the reader

more insight into the exact validation-set. For tests that originally were conducted by clients, this de-tailed description is not printed because of condentiality reasons; no extra information with respect towhat is offered in this report can be supplied.

2.1 Tests overview

All experiments that have been used, are listed in the tables below. The total experimental validation-setis divided into different categories. Each table represents a different category. The ID of the test includea reference to the category:H = head component test,S = spine component test,P = pelvis component test,F = full dummy test,R = rigid seat (full dummy) test.

Table 2.1 head tests

ID Description Conditions #F #P #V #A #IH1 Head drop certication test Standard 1

Table 2.2 spine tests

ID Description Conditions #F #P #V #A #IS1 Calibration sled test, lower/mid-spine locked 4.25m/s 3 3S2 Calibration sled test, lower/mid-spine locked 4.25m/s 3 3S3 Calibration sled test, lower/mid-spine locked 4.25m/s 3 3S4 Calibration sled test, lower/mid-spine locked

with muscle substitutes4.25m/s 3 3

S5 Calibration sled test, lower/mid-spine lockedwith muscle substitutes

4.25m/s 3 3

S6 Calibration sled test, lower/mid-spine lockedwith muscle substitutes

4.25m/s 3 3

S7 Calibration sled test, lower/mid-spine locked,damper and muscle substitutes

4.25m/s 3 3

3


10/87

Experiments MADYMO Quality Report QBioRIDII-120531

Table 2.2 spine tests

ID Description Conditions #F #P #V #A #IS8 Calibration sled test, lower/mid-spine locked,

damper and muscle substitutes4.25m/s 3 3

S9 Calibration sled test, lower/mid-spine locked,damper and muscle substitutes

4.25m/s 3 3

S10 Calibration sled test, lower spine locked, damperand muscle substitutes

4.25m/s 3 3


4.25m/s 3 3


4.25m/s 3 3

S13 Calibration sled test, whole spine, damper andmuscle substitutes

4.25m/s 3 3


4.25m/s 3 3


4.25m/s 3 3

S16 BioRID-II Calibration Test, performed by Denton mass 33.4kg, velocity4.76m/s

3 3


3 3


3 3

S19 BioRID-II Calibration Test, performed byCompany B

mass 33.4kg, velocity4.76m/s

3 3

Table 2.3 pelvis tests

ID Description Conditions #F #P #V #A #IP1 Pelvis rear-side compression test quasi-static 1 1P2 Pelvis rear-side compression test velocity 1.5m/s 1 1 1P3 Pelvis rear-side compression test velocity 2.0m/s 1 1 1P4 Pelvis rear-side compression test velocity 3.0m/s 1 1 1P5 Pelvis 20 degrees rear-side compression test velocity 2.0m/s 1 1 1P6 Pelvis 40 degrees rear-side compression test velocity 2.0m/s 1 1 1P7 Pelvis 40 degrees lower-side compression test velocity 2.0m/s 1 1 1P8 Pelvis 20 degrees lower-side compression test velocity 2.0m/s 1 1 1P9 Pelvis lower-side compression test quasi-static 1 1P10 Pelvis lower-side compression test velocity 1.5m/s 1 1 1P11 Pelvis lower-side compression test velocity 2.0m/s 1 1 1P12 Pelvis lower-side compression test velocity 3.0m/s 1 1 1

Table 2.4 full tests

ID Description Conditions #F #P #V #A #IF1 Full sled test, active head restraint present Velocity 4.4m/s 7 2F2 Full sled test, active head restraint present Velocity 4.4m/s 7 2

4


11/87


Table 2.4 full tests

ID Description Conditions #F #P #V #A #IF3 Full sled test, active head restraint present Velocity 4.4m/s 7 2

Table 2.5 rigid seat tests

ID Description Conditions #F #P #V #A #IR1 Rigid seat test, no headrest, 8km/h Velocity 2.2m/s 12 6R2 Rigid seat test, no headrest, 16km/h Velocity 4.4m/s 12 6R3 Rigid seat test, no headrest, 6km/h Velocity 1.7m/s 11 6R4 Rigid seat test, padded headrest, 12km/h Velocity 3.3m/s 11 6R5 Rigid seat test, rigid headrest, 12km/h Velocity 3.3m/s 11 6

2.2 Descriptions of the tests

2.2.1 Head Component Tests

H1 The head drop certication test conforms to the specications as written in the United States Codeof Federal Regulation Title 49 Part 572 paragraph 32. This test evaluates the contact characteristicsof the head. The results are the average of 50 performed tests.

2.2.2 Spine Component Tests

S1 Modied BioRID-II calibration test. The head and spine and torso are attached to a standard Dentoncalibration rig. The test is purely of the spine joints; the jacket, the damper and the muscle sub-stitutes are all absent from the experiment. Due to the lack of support provided to the spine, thependulum is dropped from a lower height than in the full test - 55" instead of 70". The test setupis shown in Figure 2.1 (l). The adhesive tape is present to hold the spine upright. The spine joints below the 6th thoracic vertebra were locked with two rigid steel plates. Figure 2.1 (r) shows thelower-spine restraint system.

5


12/87


Figure 2.1 Plates used to immobilise the lower spine for testing. (l); Detail of plates. (r)

S2 Modied BioRID-II calibration test. The head and spine and torso are attached to a standard Dentoncalibration rig. The test is purely of the spine joints; the jacket, the damper and the muscle sub-stitutes are all absent from the experiment. Due to the lack of support provided to the spine, thependulum is dropped from a lower height than in the full test - 55" instead of 70". The test setupis shown in Figure 2.1 (l). The adhesive tape is present to hold the spine upright. The spine joints below the 6th thoracic vertebra were locked with two rigid steel plates. Figure 2.1 (r) shows thelower-spine restraint system.

S3 Modied BioRID-II calibration test. The head and spine and torso are attached to a standard Dentoncalibration rig. The test is purely of the spine joints; the jacket, the damper and the muscle sub-stitutes are all absent from the experiment. Due to the lack of support provided to the spine, thependulum is dropped from a lower height than in the full test - 50" instead of 70". The spine joints below the 6th thoracic vertebra were locked with two rigid steel plates.

S4 Modied BioRID-II calibration test. The head and spine and torso are attached to a standard Denton

calibration rig. Neither the jacket nor the damper are included in the experiment. Due to the lackof support provided to the spine, the pendulum is dropped from a lower height than in the fulltest - 55" instead of 70". The spine joints below the 6th thoracic vertebra were locked with tworigid steel plates.

S5 Modied BioRID-II calibration test. The head and spine and torso are attached to a standard Dentoncalibration rig. Neither the jacket nor the damper are included in the experiment. Due to the lackof support provided to the spine, the pendulum is dropped from a lower height than in the fulltest - 55" instead of 70". The spine joints below the 6th thoracic vertebra were locked with tworigid steel plates.

S6 Modied BioRID-II calibration test. The head and spine and torso are attached to a standard Dentoncalibration rig. Neither the jacket nor the damper are included in the experiment. Due to the lackof support provided to the spine, the pendulum is dropped from a lower height than in the full

6


13/87


test - 50" instead of 70". The spine joints below the 6th thoracic vertebra were locked with tworigid steel plates.

S7 Modied BioRID-II calibration test. The head and spine and torso are attached to a standard Dentoncalibration rig. The jacket is not included in the experiment. Due to the lack of support providedto the spine, the pendulum is dropped from a lower height than in the full test - 55" instead of 70".The spine joints below the 6th thoracic vertebra were locked with two rigid steel plates.



S10 Modied BioRID-II calibration test. The head andspine and torso are attached to a standard Dentoncalibration rig. The jacket is not included in the experiment. Due to the lack of support providedto the spine, the pendulum is dropped from a lower height than in the full test - 55" instead of 70".The spine joints below the 12th thoracic vertebra were locked with two rigid steel plates.

S11 Modied BioRID-II calibration test. The head andspine and torso are attached to a standard Dentoncalibration rig. The jacket is not included in the experiment. Due to the lack of support providedto the spine, the pendulum is dropped from a lower height than in the full test - 55" instead of 70".The spine joints below the 12th thoracic vertebra were locked with two rigid steel plates.

S12 Modied BioRID-II calibration test. The head andspine and torso are attached to a standard Denton

calibration rig. The jacket is not included in the experiment. Due to the lack of support providedto the spine, the pendulum is dropped from a lower height than in the full test - 50" instead of 70".The spine joints below the 12th thoracic vertebra were locked with two rigid steel plates.

S13 Modied BioRID-II calibration test. The head andspine and torso are attached to a standard Dentoncalibration rig. The jacket is not included in the experiment. Due to the lack of support providedto the spine, the pendulum is dropped from a lower height than in the full test - 55" instead of 70". A sample of jacket rubber is placed between the spine and the spine support plate, to reduceexperimental noise and protect the delicate components. The test is shown in Figure 2.2

7


14/87


Figure 2.2 Dummy sub-assembly without jacket mounted on rig before testing.

S14 Modied BioRID-II calibration test. The head andspine and torso are attached to a standardDentoncalibration rig. The jacket is not included in the experiment. Due to the lack of support providedto the spine, the pendulum is dropped from a lower height than in the full test - 55" instead of 70". A sample of jacket rubber is placed between the spine and the spine support plate, to reduceexperimental noise and protect the delicate components. The test is shown in Figure 2.2

S15 Modied BioRID-II calibration test. The head andspine and torso are attached to a standardDentoncalibration rig. The jacket is not included in the experiment. Due to the lack of support providedto the spine, the pendulum is dropped from a lower height than in the full test - 50" instead of 70". A sample of jacket rubber is placed between the spine and the spine support plate, to reduceexperimental noise and protect the delicate components.

S16 Standard BioRID-II calibration test. The head, spine and torso are attached to a specially-designedrig, shown in Figure 2.3. The base of the spine is held in place and the back rests against a curvedsteel plate. A 33.4kg pendulum strikes the back of the rig with an impact velocity of 4.76 m/s,

accelerations and rotations are measured and compared to calibration corridors.

8


15/87


Figure 2.3 Standard BioRID-II calibration test.

S17 Standard BioRID-II calibration test. The head, spine and torso are attached to a specially-designed

rig, shown in Figure 2.3. The base of the spine is held in place and the back rests against a curvedsteel plate. A 33.4kg pendulum strikes the back of the rig with an impact velocity of 4.76 m/s,accelerations and rotations are measured and compared to calibration corridors.

S18 Standard BioRID-II calibration test. The head, spine and torso are attached to a specially-designedrig, shown in Figure 2.3. The base of the spine is held in place and the back rests against a curvedsteel plate. A 33.4kg pendulum strikes the back of the rig with an impact velocity of 4.76 m/s,accelerations and rotations are measured and compared to calibration corridors.

S19 Standard BioRID-II calibration test. The head, spine and torso are attached to a specially-designedrig. The base of the spine is held in place and the back rests against a curved steel plate. A33.4kg pendulum strikes the back of the rig with an impact velocity of 4.76 m/s, accelerations and

rotations are measured and compared to calibration corridors.

2.2.3 Pelvis Component Tests

P1 This is a quasi-static pelvis lower-side compression test with a 300X300mm rigid plate. The pelviswas supported on a rigid frame via the lumbar spine bracket. The impactor plate, attached to anhydraulic piston, loads and unloads the pelvis at 0.001m/s velocity. The impactor displacementand force were measured. Figure 2.4 shows the experimental test setup.

9


16/87


Figure 2.4 Quasi-static pelvis rear-side compression test at t=0

P2 This is a dynamic pelvis rear-side compression test with a 300X300mm rigid plate. The pelvis was

supported on a rigid frame via the lumbar spine bracket. At the time of impact the plate hitsthe pelvis at 1.5m/s velocity. The impactor displacement, acceleration and force were measured.Figure 2.5 shows the experimental test setup.

Figure 2.5 Dynamic pelvis rear-side compression test at t=0

P3 This is a dynamic pelvis rear-side compression test with a 300X300mm rigid plate. The pelvis wassupported on a rigid frame via the lumbar spine bracket. At the time of impact the plate hitsthe pelvis at 2.0m/s velocity. The impactor displacement, acceleration and force were measured.Figure 2.5 shows the experimental test setup.

P4 This is a dynamic pelvis rear-side compression test with a 300X300mm rigid plate. The pelvis wassupported on a rigid frame via the lumbar spine bracket. At the time of impact the plate hitsthe pelvis at 3.0m/s velocity. The impactor displacement, acceleration and force were measured.

10


17/87


Figure 2.5 shows the experimental test setup.

P5 This is a dynamic pelvis 20 degrees rear-side compression test with a 300X300mm rigid plate. Thepelvis was supported on a rigid frame via the lumbar spine bracket. At the time of impact theplate hits the pelvis at 2.0m/s velocity. The impactor displacement, acceleration and force weremeasured. Figure 2.6 shows the experimental test setup.

Figure 2.6 Dynamic pelvis rear-side compression test at t=0

P6 This is a dynamic pelvis 40 degrees rear-side compression test with a 300X300mm rigid plate. Thepelvis was supported on a rigid frame via the lumbar spine bracket. At the time of impact theplate hits the pelvis at 2.0m/s velocity. The impactor displacement, acceleration and force weremeasured. Figure 2.7 shows the experimental test setup.

Figure 2.7 Dynamic pelvis 40 degrees rear-side compression test at t=0

P7 This is a dynamic pelvis 40 degrees lower-side compression test with a 300X300mm rigid plate. Thepelvis was supported on a rigid frame via the lumbar spine bracket. At the time of impact theplate hits the pelvis at 2.0m/s velocity. The impactor displacement, acceleration and force weremeasured. Figure 2.8 shows the experimental test setup.

11


18/87


Figure 2.8 Dynamic pelvis 40 degrees lower-side compression test at t=0

P8 This is a dynamic pelvis 20 degrees lower-side compression test with a 300X300mm rigid plate. Thepelvis was supported on a rigid frame via the lumbar spine bracket. At the time of impact theplate hits the pelvis at 2.0m/s velocity. The impactor displacement, acceleration and force weremeasured. Figure 2.9 shows the experimental test setup.

Figure 2.9 Dynamic pelvis 40 degrees lower-side compression test at t=0

P9 This is a quasi-static pelvis lower-side compression test with a 300X300mm rigid plate. The pelviswas supported on a rigid frame via the lumbar spine bracket. The impactor plate, attached to anhydraulic piston, loads and unloads the pelvis at 0.001m/s velocity. The impactor displacementand force were measured. Figure 2.10 shows the experimental test setup.

12


19/87


Figure 2.10 Quasi-static pelvis lower-side compression test at t=0

P10 This is a dynamic pelvis lower-side compression test with a 300X300mm rigid plate. The pelvis

was supported on a rigid frame via the lumbar spine bracket. At the time of impact the plate hitsthe pelvis at 1.5m/s velocity. The impactor displacement, acceleration and force were measured.Figure 2.11 shows the experimental test setup.

13


20/87


Figure 2.11 Dynamic pelvis lower-side compression test at t=0

P11 This is a dynamic pelvis lower-side compression test with a 300X300mm rigid plate. The pelvis

was supported on a rigid frame via the lumbar spine bracket. At the time of impact the plate hitsthe pelvis at 2.0m/s velocity. The impactor displacement, acceleration and force were measured.Figure 2.11 shows the experimental test setup.

P12 This is a dynamic pelvis lower-side compression test with a 300X300mm rigid plate. The pelviswas supported on a rigid frame via the lumbar spine bracket. At the time of impact the plate hitsthe pelvis at 3.0m/s velocity. The impactor displacement, acceleration and force were measured.Figure 2.11 shows the experimental test setup.

2.2.4 Full Dummy Tests

F1 Full dummy sled test, 16km/h, peak acceleration 10g. This test was carried out for a client, as thedetails are condential.



2.2.5 Rigid Seat Tests

R1 Full dummy rigid-seat sled test, 8km/h. The dummy was placed in a sled-mounted rigid seatdesigned by JARI. No head restraint was present, and a rear impact pulse giving a delta-V of

8km/h was applied to the sled. The seatback was set at 20 degrees and the dummy was placed

14


21/87


in it, following standard placement protocol. Head, spine and pelvis accelerations were measuredusing accelerometers; upper and lower neck loads were measured with load cells. In addition, arecording of the test was made using two high-speed video cameras. This experiment was carriedout at JARI.

R2 Full dummy rigid-seat sled test, 16km/h. The dummy was placed in a sled-mounted rigid seatdesigned by JARI. A head restraint was present to protect the dummy neck, and a rear impactpulse giving a delta-V of 16km/h was applied to the sled. The seatback was set at 20 degreesand the dummy was placed in it, following standard placement protocol. Head, spine and pelvisaccelerations were measured using accelerometers; upper and lower neck loads were measuredwith load cells. In addition, a recording of the test was made using two high-speed video cameras.This experiment was carried out at JARI.

R3 Full dummy rigid-seat sled test, 6km/h. The dummy was placed in a sled-mounted rigid seat withno head restraint present and a mild pulse giving a delta-V of 6km/h was applied to the sled. Thedummy was placed in the seat following standard placement protocol. Head, spine and pelvisaccelerations were measured using accelerometers; upper and lower neck loads were measuredwith load cells. In addition, a recording of the test was made using two high-speed video cameras.

R4 Full dummy rigid-seat sled test, 12km/h. The dummy was placed in a sled-mounted rigid seatwith a foam pad block present to act as a head restraint. A pulse giving a delta-V of 12km/h wasapplied to the sled. The dummy was placed in the seat following standard placement protocol.Head, spine and pelvis accelerations were measured using accelerometers; upper and lower neckloads were measured with load cells. In addition, a recording of the test was made using twohigh-speed video cameras.

R5 Full dummy rigid-seat sled test, 12km/h. The dummy was placed in a sled-mounted rigid seat withits head resting against a rigid head restraint. A pulse giving a delta-V of 12km/h was applied tothe sled. The dummy was placed in the seat following standard placement protocol. Head, spineand pelvis accelerations were measured using accelerometers; upper and lower neck loads weremeasured with load cells. In addition, a recording of the test was made using two high-speedvideo cameras.

15


22/87


16


23/87

MADYMO Quality Report QBioRIDII-120531 Rating of the validation set

3 Rating of the validation set

All signals of all tests (40 experiments) have been numerically rated in the way as described in theappendix. The table below lists the combined values of all tests. Note that component scores includeall tests, but only the signals that are measured in the component are included in the calculation of the

score. Combined signals are also given.

3.1 Overall test results

In this section the rating results are presented in tables. The rst four tables list the overall rating resultsfor the dummy model. The rst three give the score per rating criterion for the total dummy validationtest set, followed by those for for each test group; the fourth table gives the combined score (combiningthe scores from all three rating criteria) for the total test set, followed by those for each test group. Inthese tables, the second column shows the weight factor that has been applied to the score of each testgroup for calculating the total scores. These test group weight factors are calculated as the ratio of thenumber of tests within the test group and the number of tests in the total dummy validation test set. Inthis way the sum of the test group weight factors is always 1.0. In the third column of the tables, thescores are given in percentages, with 100% indicating a perfect match with the experimental data.

After these rst four tables, additional tables present the combined rating results of the individual testsin each test group. In each test group, all the tests (referred to by their test ID) are given the same testweight factor. The test weight factors are calculated as the inverse of the number of tests in the test groupto which it belongs. In this way, the sum of the test weight factors in a test group is always 1.0. Usingthe combination of test weight factors and test group weight factors, the score from each individual testcontributes equally to the total score for the complete dummy validation test set.

Table 3.1 Rating results for the model using the Peak criter ion only

Group Weight Model M2Total 70.4% 69.5%head tests 0.0250 93.5% 93.5%spine tests 0.4750 65.3% 63.5%pelvis tests 0.3000 86.9% 86.9%full tests 0.0750 74.2% 75.8%rigid seat tests 0.1250 59.6% 59.7%

Table 3.2 Rating results for the model using the Peaktime criterion only


17


24/87

Rating of the validation set MADYMO Quality Report QBioRIDII-120531

Table 3.3 Rating results for the model using the Wifac criterion only

Group Weight Model M2Total 50.8% 51.2%head tests 0.0250 89.0% 89.0%

spine tests 0.4750 46.0% 45.1%pelvis tests 0.3000 63.9% 63.9%full tests 0.0750 51.1% 57.7%rigid seat tests 0.1250 39.3% 42.2%

Table 3.4 Combined rating results for the model


Table 3.5 Combined rating results for the head tests

Group Weight Model M2H1 1.0000 92.6% 92.6%

Table 3.6 Combined rating results for the spine tests

Group Weight Model M2S1 0.0526 60.0% 60.0%S2 0.0526 57.6% 57.6%S3 0.0526 58.4% 58.4%S4 0.0526 61.7% 61.7%S5 0.0526 59.5% 59.5%S6 0.0526 59.2% 59.2%S7 0.0526 60.8% 60.8%S8 0.0526 61.8% 61.8%S9 0.0526 62.0% 62.0%S10 0.0526 62.0% 62.0%S11 0.0526 59.4% 59.4%S12 0.0526 59.8% 59.8%S13 0.0526 67.6% 67.6%S14 0.0526 65.7% 65.7%S15 0.0526 67.0% 67.0%S16 0.0526 60.3% 60.3%S17 0.0526 59.1% 59.0%S18 0.0526 57.9% 60.3%S19 0.0526 57.7% 38.1%

18


25/87

MADYMO Quality Report QBioRIDII-120531 Rating of the validation set

Table 3.7 Combined rating results for the pelvis tests

Group Weight Model M2P1 0.0833 87.5% 87.5%P2 0.0833 76.8% 76.8%

P3 0.0833 71.7% 71.7%P4 0.0833 67.9% 67.9%P5 0.0833 77.6% 77.6%P6 0.0833 81.1% 81.1%P7 0.0833 77.9% 77.9%P8 0.0833 79.1% 79.1%P9 0.0833 89.2% 89.2%P10 0.0833 71.7% 71.7%P11 0.0833 79.2% 79.2%P12 0.0833 68.9% 68.9%

Table 3.8 Combined rating results for the full tests

Group Weight Model M2F1 0.3333 68.6% 68.5%F2 0.3333 68.1% 70.5%F3 0.3333 65.9% 74.0%

Table 3.9 Combined rating results for the rigid seat tests

Group Weight Model M2R1 0.2000 55.8% 62.5%R2 0.2000 51.6% 53.9%R3 0.2000 62.7% 64.5%R4 0.2000 61.3% 62.9%R5 0.2000 50.4% 40.5%

19


26/87

Rating of the validation set MADYMO Quality Report QBioRIDII-120531

20


27/87

MADYMO Quality Report QBioRIDII-120531 Comparison of results

4 Comparison of results

This chapter shows the results that are obtained directly from the experiment and simulations. Directsignal output is listed in an Appendix. Range plots and Trend plots are shown in the next paragraphs.The next chapter shows rating results that are obtained from numerical comparisons between signals.

4.1 Range plots

The range plots show the results of a particular signal over different tests. The peak value of the signalduring a particular simulation, is represented by a dot in the graph. The horizontal location of the dotis proportional to the experimental test signal peak. In general this corresponds to the test severity. Thevertical position is proportional to the simulation peak results. If the simulation corresponds with theexperiment, the dot is on the 45 degree line.If the dot is below this line, the simulation had a lower peak than the experiment. Next to the 45 degreeline, two lines have been drawn. Within these lines the peak score is above 0.8.

0

1000

2000

3000

0 1000 2000 3000

S i m u l a t i o n

Experiment

-200

0

200

400

-200 0 200 400

S i m u l a t i o n

Experiment

Figure 4.1 Rangeplot of signal Head_AccR (left); Rangeplot of signal Head_AccX (right)

-200

0

200

-200 0 200

S i m u l a t i o n

Experiment

-0.8

-0.4

0

0.4

-0.8 -0.4 0 0.4

S i m u l a t i o n

Experiment

Figure 4.2 Rangeplot of signal Head_AccZ (left); Rangeplot of signal Head_AdisY (right)

21


28/87

Comparison of results MADYMO Quality Report QBioRIDII-120531

-1.2

-0.8

-0.4

0

0.4

-1.2 -0.8 -0.4 0 0.4

S i m u l a t i o n

Experiment

-200

0

200

-200 0 200

S i m u l a t i o n

Experiment

Figure 4.3 Rangeplot of signal NeckLink_AdisY (left); Rangeplot of signal T1_AccX (right)

-0.4

-0.2

0

0.2

-0.4 -0.2 0 0.2

S i m u l a t i o n

Experiment

-20000

-15000

-10000

-5000

0

-20000 -15000 -10000 -5000 0

S i m u l a t i o n

Experiment

Figure 4.4 Rangeplot of signal T1_AdisY (left); Rangeplot of signal Pelvis_FrcX (right)

-0.06

-0.04

-0.02

0

0.02

-0.06 -0.04 -0.02 0 0.02

S i m u l a t i o n

Experiment

-800

-400

0

400

-800 -400 0 400

S i m u l a t i o n

Experiment

Figure 4.5 Rangeplot of signal Impactor_DisX (left); Rangeplot of signal Impactor_AccX (right)

22


29/87


-4000

-2000

0

-4000 -2000 0

S i m u l a t i o n

Experiment

-20000

-15000

-10000

-5000

0

-20000 -15000 -10000 -5000 0

S i m u l a t i o n

Experiment

Figure 4.6 Rangeplot of signal Impactor_FrcX (left); Rangeplot of signal Pelvis_FrcZ (right)

-200

0

200

-200 0 200

S i m u l a t i o n

Experiment

-100

0

100

200

-100 0 100 200

S i m u l a t i o n

Experiment

Figure 4.7 Rangeplot of signal Pelvis_AccX (left); Rangeplot of signal Pelvis_AccZ (right)

0

400

800

1200

1600

0 400 800 1200 1600

S i m u l a t i o n

Experiment

-500

0

500

1000

1500

-500 0 500 1000 1500

S i m u l a t i o n

Experiment

Figure 4.8 Rangeplot of signal NeckUp_FrcX (left); Rangeplot of signal NeckUp_FrcZ (right)

23


30/87


31/87


-40

-20

0

20

-40 -20 0 20

S i m u l a t i o n

Experiment

-100

0

100

200

-100 0 100 200

S i m u l a t i o n

Experiment

Figure 4.12 Rangeplot of signal NeckLow_MomY (left); Rangeplot of signal C4_AccZ (right)

-100

0

100

200

-100 0 100 200

S i m u l a t i o n

Experiment

-100

0

100

200

-100 0 100 200

S i m u l a t i o n

Experiment

Figure 4.13 Rangeplot of signal T1_AccZ (left); Rangeplot of signal T8_AccZ (right)

-100

0

100

200

-100 0 100 200

S i m u l a t i o n

Experiment

Figure 4.14 Rangeplot of signal L1_AccZ

4.2 CPU Time comparison

This section presents information about the runtimes of the simulations. From this information, thereader can get an indication of the CPU demands made by typical applications.

For the results presented in this section the following machine/platform was used: 2.5GHz Quad-CoreAMD Opteron, 1GHz FSB with Linux operating system.

25


32/87

Comparison of results MADYMO Quality Report QBioRIDII-120531

The CPU time for the head test is around a few seconds. The pelvis component tests take between 45seconds and two minutes, and spine component tests between 1 and 4 minutes, with the certicationtests using the FE jacket taking around 3.5 hours.

The full dummy tests with rigid seats take between 15 and 30 minutes to run for the facet model, and

around 7 hours for the FE model.The full dummy tests with exible facet seats take around 25 minutesto run for the facet model, and around 3.5 hours for the FE model.

26


33/87

MADYMO Quality Report QBioRIDII-120531 Rating Method

A Rating Method

A.1 Introduction

This section describes the algorithms used to obtain the rating values for models, and also discussestheir backgrounds.

A number of changes in the algorithms has been implemented per July 1, 2006, and scores in newerreports may be different from those presented in older reports. The changes have been made in order tosolve a number of small bugs in the code, treat 0 values properly, and improve WIFac scores for signalsthat (repeatedly) cross the abscissa.

The rst change is in the way that the comparison of the scalar value 0 is handled. All algorithms usedto calculate scores will now rst compare the two scalars and return a value of 1 if they are equal. Thismeans that if both values to be compared are zero the rating will be 1 (100%). The previous version of

the rating algorithm would return a zero (0% correlation). This change applies to calculations of peakamplitude, timing of the peak amplitude and WIFac calculations. The second change is in the WIFacMethod where comparing signals values around zero use lower weighing factors than values aroundpeaks. This has been achieved by squaring the weight factors instead of using the linear weighting usedin the older code.

A.1.1 Rating a MADYMO dummy model

MADYMO expresses the quality of a dummy model in a single number. In order to derive the ratingfor a dummy model simulation outputs are compared to experimental data. Each experiment on adummy, or a part of a dummy, will result in a number of signals, e.g. head accelerations, neck forces,etc. Corresponding simulation time history data can be generated by modelling the experiment. Criteriathat allow a quantative comparison between two signals can be used to derive a score for each signal.For example, the peak value of a signal is almost always used to compare and rate simulation output.More than one comparison can be applied to each signal. Individual scores are then added in a denedway to derive a total score for the model.

In MADYMO dummy model development, the complete set of experiments to be used to rate a modelis divided into several groups and sub-groups. The smallest entity is just a single test, for example ahead drop test performed according to the part 572 protocol. Applying rating calculation to the resultsof this test can result in up to three values: one for the peak value, one for the timing of the peak value,and one for the shape of the signal.

Several similar tests can be grouped together and rated as a group, in which a weighted average isdetermined of the individual scores. To continue the example, a series of tests according to the part 572protocol can be applied to the head at different velocities and an average score can be calculated fromthe individual scores.

On the next level, the rating scores of groups of dissimilar tests can be combined to derive a total scorefor a certain part or component. For example, several types of dynamic tests can be combined with statictests on dummy components or sub-assemblies. The purpose of the grouping is to allow the user of themodel to look not only at the total dummy rating score, but also to examine certain aspects of the modelin greater detail.

On the highest level, the MADYMO rating system uses three categories of groups, of which the scoresare combined in a single scalar value. These categories are:

27


34/87

Rating Method MADYMO Quality Report QBioRIDII-120531

1. Full: Full tests are typically of the complete dummy in a test environment. Examples of full testsare NCAP tests or other complete vehicle tests and sled tests. In simulations of this type not onlythe dummy, but also the environment needs to be accurately modelled. Simulations of the fulltype are used for model validation.

2. Assembly: An assembly test is typically not on a complete dummy, but on a subset of the dummy,and is intended to study the interaction between dummy parts under dynamic conditions.

3. Part: These are tests on isolated parts of the dummy, for example the head, neck or pelvis. Thepurpose of the test is to determine the dynamic properties of the part. Examples of tests in thiscategory are all dummy certication tests, such as the part 572 head drop test or the EuroSIDdummy neck certication test.

As previously mentioned, signals can be compared in several ways to derive multiple scores describingthe correlation between experimental data and simulation data. This comparison can either use scalars,for example comparing peak signals values or the HIC calculated from the head accelerations, or vectors,for example by calculating a phase shift between two signals or comparing the areas under the curves.

How these comparisons are performed is described in the next two paragraphs. Additional backgroundinformation on comparing signals is given at the end of this appendix.

A.1.2 Comparing two scalar values

To calculate the scores between two scalar values, for example peak values or injury values, MADYMOuses the following equation, referred to as the Factor Method (FM):

crit = max (0,exp sim )

max (exp 2,sim 2) (A.1)

Where crit is a derived scalar (score), and exp and sim are two scalars describing an aspect of the exper-

imental and simulation data. When both values, exp and sim, are zero the result of crit will be set to 1(100%).The FM calculates the correlation between exp and sim, resulting in a value that lies between 0 and 1.A characteristic of the FM is that it calculates values that do not drop off faster for overprediction thanfor underprediction. This means that using FM to optimise models does not result in models that eithersystematically under- or overpredict dummy response.

A.1.3 Comparing two signals

Currently, one method is available to calculate the correlation between experimental and simulationdata that uses entire time histories. It is called the WIFac Method , which stands for Weighted IntegratedFactor method. Its formulation is as follows:

crit = 1 n (max (f [n ]2, g [n ]2) (1 max (0 ,f [n ]g[n ])max (f [n ]2 ,g [n ]2 ) )2)n max (f [n ]2, g[n ]2)

(A.2)

Where f[n] and g[n] are the two signals that are to be compared over a time interval n. When bothvalues, f[n] and g[n], are zero the result of crit will be set to 1 (100%). Equation ( A.2) is the combinationof Equation ( A.1) for the calculation of a score from a single scalar with the following equation forcalculating a weighted average from a series of scalar scores.

score = 1

n (W [n ] (1 crit [n ])2)

n W [n ]

(A.3)

Where W are weights, and crit are rating scores for each set of data points.

28


35/87


In Equation ( A.2), the weight factor W is calculated as the maximum of the square of the two functionvalues at each point, and Equation ( A.1) is used to calculate crit for each timestep n. Effectively thismeans that every individual score is scaled such that it contributes to the total score just as the functionvalue would contribute to the total area underneath the graph. In other words, low signal values havelower weights than high signal levels.

A.1.4 Adding scores

The previous paragraphs have described how data from experimental and simulation data can be com-pared and how the correlation between the two signals can be expressed with a single scalar. Multiplecomparisons can be made between any set of two time history signals resulting in a set of scalars.To obtain a single rating number for the complete model or model component, results from differenttests and signals need to be combined. For reasons of comparison, the domain of the combined numberis often taken identical to the domain of the individual numbers. Therefore for combined numbers thedomain [0, 1] is used. Just as for the individual numbers, a score of 0 means that there is no correlationat all, and a score of 1 (100%) means a perfect match.The equation used to calculate weighted averages is called the Root-Mean-Square Method (RMS):

comb = 1 (W (1 crit )2)W (A.4)Where comb is a combined score, W are weights for scores, and crit are rating scores.Equation ( A.4) is used on all levels to calculate averaged scores, resulting in a single scalar that expressesthe quality of the model. On the lowest level, it is used to calculate the avarage value of the scores forpeak amplitude, timing of the peak amplitude, and the WIFac for a single signal. On the next level,scores for all signals are averaged to calculate a score for a single simulation/test. One level up, anaverage score is calculated for all similar tests, etc, etc.

A.2 Additional information on calculation of dummy rating scoresThis section will give some background information on the methods used to calculate the scores andthen illustrate the differences with an example.

A.2.1 Comparing two scalar values

To calculate the scores between two scalar values, for example peak values or injury values, the follow-ing expression is often used:

crit = max (0, 1 |exp sim |

|exp | ) (A.5)

This method will be called the Relative Error Method (REM). The REM is not associative: crit(exp, sim)= crit(sim, exp). The REM score drops off faster with overprediction than with underprediction. Asa result, applying this formula to optimize a model to a set of tests, tends to result in a model thatsystematicaly underpredicts test outcomes. Since this underprediction is undesirable for the dummymodels the following calculation is used instead by MADYMO, which is called the Factor Method:

crit = max (0,exp sim )

max (exp 2,sim 2) (A.6)

When both values, f[n] and g[n], are zero the result of crit will be set to 1 (100%). The Factor Method(FM) calculates the correlation between exp and sim, resulting in a value that lies between 0 and 1. Thisexpression is associative: crit(exp, sim) = crit(sim,exp). For small differences the results are identical oralmost identical to the results of the REM. However, it is important to realise that for overpredictions theFM calculates values that do not drop off faster than for underprediction, as is the case with the REM.

29


36/87


For example for an overprediction of 100%, using FM results in a score of 0.5, whereas the REM scorewill have dropped to 0. When optimising a target function using the FM, the solution will not tend tounderpredict most tests as it would when the REM had been used.

A.2.2 Adding scores

To obtain a single rating number for the complete model or model component, results from differenttests and signals need to be combined. For reasons of comparison, the domain of the combined numberis often taken identical to the domain of the individual numbers. Therefore for combined numbers alsothe domain [0, 1] (x100%) is used. Just as for the individual numbers, a score of 0 means that thereis no correlation at all, and a score of 1 (100%) designates a perfect match. Sometimes the followingexpression is used to achieve a combined score, which is called the Simple Addition Method (SAM) inthis document:

comb =(W crit )

W (A.7)

Where comb is a combined score, W are weights for scores, and crit are rating scores.When a model is optimised using the maximum value of the combined score according to Equation(A.7), it might very well end in a point where it has a maximum score for one test, and a low score forall other tests. For this type of optimal solution, the solution is not sensitive to small variations in all butone or some tests in specic. The value of the Simple Addition Method will vary when one of the lessimportant tests varies, but the optimal design point will not be adjusted in all cases.The risk of an optimal solution that lacks sensitivity for the results of many tests, is the main disadvan-tage of the Simple Addition Method. Therefore within MADYMO dummy development the Root MeanSquare (RMS) addition is used (Equation ( A.4)). When this formula is used in an optimisation process,the optimal design point will always vary when the result of one of the tests varies. Furthermore op-timising the total score calculated with the RMS Addition method is equal to optimising the problemwith the generally known least squares method (Equation (A.4)).

A.2.3 Example

In this example a model is optimised using two tests. The model is optimised by varying a singleparameter K, which represents some parameter of the model. In dummy models, K can for example be stiffness or damping. Most often it is not possible to nd a value for the parameter that results in aperfect correlation for all tests. In this example it is assumed that K scales the response of the model inthe following (conicting) way:sim1(K) = 0.5/K, sim2(K) = 1.0/K.The scores of sim1 and sim2 are relative to the goal value of 1. So there are two values of K for whichone of the two tests correlates perfectly:

K=0.5: sim1 = 1 .0 exp, a perfect match for test 1.

sim2 = 2 .0 exp, an overprediction of 100% for test 2.

K=1.0: sim1 = 0 .5 exp, an underprediction of 50% for test 1.

sim2 = 1 .0 exp, a perfect match for test 2.

Rating values calculated using the Relative Error Method (Equation ( A.5)) are shown in Figure A.1.

30


37/87


0

0.2

0.4

0.6

0.8

1

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

C r i t

K

Test 1Test 2

0

0.2

0.4

0.6

0.8

1

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

C r i t

K

Test 1Test 2

Figure A.1 Rating values calculated using the Relative Error Method (l) and the Factor Method (r).

Figure A.1 shows a much steeper drop-off for overprediction than for underprediction. Rating valuescalculated using the Factor Method (Equation 2) are shown in Figure A.1. Here the drop-off for under-

and overprediction is about equal within a certain range from the peak value.The two scores of the tests can be combined using the Simple Addition Method (Equation ( A.7)), andRMS Addition Method (Equation ( A.4)). The results for the Simple Addition Method are shown inFigure A.2.

0

0.2

0.4

0.6

0.8

1

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

a v e r a g e

K

REMFM

0

0.2

0.4

0.6

0.8

1

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

a v e r a g e

K

REMFM

Figure A.2 SAM (l) and RMS (r) using Relative Error and Factor Method.

For the Relative Error Method Figure A.2 shows a maximum average score of 0.75 at K = 1.0. For thisvalue of K the performance for test 2 is optimal, but the result of test 1 is underpredicted by 50%. The factthat the optimum value gives a perfect match for one test, and a poor for another is because the Simple

Addition Method uses linear terms in its calculation, and the REM drop-off is not symmetric for under-and overprediction. In this example, the optimum is found at that test that requires the highest value of K. Other tests will be underpredicted. Using the Factor Method two maximum values are found, onefor K = 0.5 and one for K = 1.0. Two maxima are found because the Factor Method is symmetric for over-and underprediction. Whichever method is used to calculate the single rating scores, using the SimpleAddition Method results in a optimum that has a high score for one test and a poor score for the othertest. However, in model optimization, a solution is sought that nds a balance between the two tests,i.e. that the results of both tests are as close as possible to the required values. The results for additionof scores with the RMS Method are shown in Figure A.2.

Because the RMS addition does not use linear terms in its calculations, none of the tests in this example

can dominate the calculation of the optimum value. For both the Relative Error Method and the FactorMethod, the optimal value for K lies between 0.5 and 1.0. If used with the Factor Method ( = WIFac), theoptimum value is K = 0.707. If used with the Relative Error Method, the optimum value is much closer

31


38/87


to K = 1.0, because the Relative Error Method is in favor of underprediction.

The Tables A.1 and A.2 summarize the results for Simple Addition and Root-Mean-Square.

REM FM

Test 1 0.5 1.0 2.0Test 2 1.0 2.0 1.0

Table A.1 SAM

REM FMTest 1 0.588235 0.707214Test 2 1.176471 1.414427

Table A.2 RMS

A.3 Results for more complex examples

The previous example showed that for a model with two tests and one model parameter that scalesthe test results, the Factor Method combined with RMS Addition gives much better results than thecombination of the Relative Error Method and Simple Addition. However, the question is if this is stillthe case for a model with more parameters and more tests. In this section the results are shown for amodel with a single parameter K and 6 tests. The optimal values for K are chosen as 0.5, 0.6, 0.7, 0.8, 0.9and 1.0. Only the Factor Method criterion has been used during this evaluation.

0

0.2

0.4

0.6

0.8

1

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

a v e r a g e

K

SAMRMS

0

0.2

0.4

0.6

0.8

1

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

a v e r a g e

K

SAMRMS

Figure A.3 SAM and RMS for 6 tests with equal weight factors (l) and with unequal weight factors (r).

For this case the Simple Addition curve is much more uent than in the case for two tests. Howeverthere are still peaks at every test. The optimum value for the Simple Addition is 0.8, which is exactlyat one of the tests (K=0.8). Another peak is at 0.7, but its value is slightly lower. The reason that thesevalues are not equal is that the distribution of the optimal K values between 0.5 and 1.0 was chosenlinear, not harmonic. If one of the tests has a weighting factor such that it counts more in the additionthan the other tests, the results for the Simple Addition Method are not optimal, as is shown in FigureA.3.In this example there are 3 tests with optimal K at 0.5, 0.7 and 1.0. The test with optimum at K = 1.0weighs just as much as the other two tests together. The optimum with the Simple Addition Methodis at K = 1.0, which more or less ignores the other tests. The RMS Addition has its optimum around K= 0.78, which is between the optimal K values for the tests. From this more complicated example it isconcluded that the Simple Addition Method can result in sub-optimal solutions. MADYMO therefore

32


39/87


uses the RMS Addition. However, it should be noted that the RMS Addition leads to a lower score thanthe Simple Addition. The exact difference depends on the case, so it can not be compensated for usinga scaling factor.

33


40/87


34


41/87


42/87

Signal results MADYMO Quality Report QBioRIDII-120531

-0.8

-0.4

0

0.4

0 100 200

H e a d

_ A d i s Y

Time [ms]

81.0% 97.4% 68.1% 78.5%81.0% 97.4% 68.1% 78.5%

-1.2

-0.8

-0.4

0

0 100 200

N e c k L i n k_ A d i s Y

Time [ms]

87.7% 96.9% 90.4% 90.8%87.7% 96.9% 90.4% 90.8%

Figure B.3 Test S1 signal Head_AdisY (l); Test S1 signal NeckLink_AdisY (r)

-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

37.2% 91.8% 26.4% 44.0%37.2% 91.8% 26.4% 44.0%

-0.4

-0.2

0

0.2

0 100 200

T 1

_ A d i s Y

Time [ms]

87.6% 85.6% 78.0% 83.2%87.6% 85.6% 78.0% 83.2%

Figure B.4 Test S1 signal T1_AccX (l); Test S1 signal T1_AdisY (r)

-100

0

100

0 100 200

H e a d

_ A c c X

Time [ms]

29.5% 98.5% 34.5% 44.4%29.5% 98.5% 34.5% 44.4%

-100

0

100

200

0 100 200

H e a d

_ A c c Z

Time [ms]

41.9% 60.2% 25.4% 40.8%41.9% 60.2% 25.4% 40.8%

Figure B.5 Test S2 signal Head_AccX (l); Test S2 signal Head_AccZ (r)

36


43/87

MADYMO Quality Report QBioRIDII-120531 Signal results

-0.8

-0.4

0

0.4

0 100 200

H e a d

_ A d i s Y

Time [ms]

84.0% 96.7% 70.1% 80.3%84.0% 96.7% 70.1% 80.3%

-1.2

-0.8

-0.4

0

0 100 200


Time [ms]

89.7% 97.9% 89.9% 91.6%89.7% 97.9% 89.9% 91.6%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

34.0% 97.6% 24.3% 42.0%34.0% 97.6% 24.3% 42.0%

-0.4

-0.2

0

0.2

0 100 200

T 1

_ A d i s Y

Time [ms]

83.5% 82.6% 78.5% 81.4%83.5% 82.6% 78.5% 81.4%


-100

0

100

0 100 200

H e a d

_ A c c X

Time [ms]

25.8% 99.0% 33.1% 42.3%25.8% 99.0% 33.1% 42.3%

-100

0

100

200

0 100 200

H e a d

_ A c c Z

Time [ms]

51.4% 60.5% 25.4% 43.8%51.4% 60.5% 25.4% 43.8%


37


44/87


-0.8

-0.4

0

0.4

0 100 200

H e a d

_ A d i s Y

Time [ms]

85.2% 90.0% 69.3% 79.5%85.2% 90.0% 69.3% 79.5%

-1.2

-0.8

-0.4

0

0 100 200


Time [ms]

87.1% 97.9% 89.1% 90.2%87.1% 97.9% 89.1% 90.2%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

36.7% 96.1% 28.7% 44.9%36.7% 96.1% 28.7% 44.9%

-0.4

-0.2

0

0.2

0 100 200

T 1

_ A d i s Y

Time [ms]

84.9% 83.6% 76.0% 81.1%84.9% 83.6% 76.0% 81.1%


-50

0

50

100

150

0 100 200

H e a d

_ A c c X

Time [ms]

49.1% 99.7% 56.9% 61.5%49.1% 99.7% 56.9% 61.5%

-100

0

100

200

0 100 200

H e a d

_ A c c Z

Time [ms]

57.7% 60.7% 25.0% 45.4%57.7% 60.7% 25.0% 45.4%


38


45/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

76.5% 70.7% 52.9% 65.2%76.5% 70.7% 52.9% 65.2%

-0.8

-0.4

0

0.4

0 100 200


Time [ms]

82.6% 96.2% 85.8% 86.8%82.6% 96.2% 85.8% 86.8%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

39.3% 87.1% 33.2% 47.3%39.3% 87.1% 33.2% 47.3%

-0.4

-0.2

0

0 100 200

T 1

_ A d i s Y

Time [ms]

89.8% 98.6% 79.5% 86.8%89.8% 98.6% 79.5% 86.8%


-50

0

50

100

150

0 100 200

H e a d

_ A c c X

Time [ms]

46.4% 78.0% 54.7% 57.5%46.4% 78.0% 54.7% 57.5%

-100

0

100

200

0 100 200

H e a d

_ A c c Z

Time [ms]

58.3% 61.2% 21.0% 43.8%58.3% 61.2% 21.0% 43.8%


39


46/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

74.6% 71.5% 50.6% 63.9%74.6% 71.5% 50.6% 63.9%

-0.8

-0.4

0

0.4

0 100 200


Time [ms]

83.3% 96.9% 85.9% 87.3%83.3% 96.9% 85.9% 87.3%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

34.6% 79.8% 29.6% 43.3%34.6% 79.8% 29.6% 43.3%

-0.4

-0.2

0

0 100 200

T 1

_ A d i s Y

Time [ms]

89.2% 94.8% 79.3% 86.2%89.2% 94.8% 79.3% 86.2%


-50

0

50

100

150

0 100 200

H e a d

_ A c c X

Time [ms]

45.7% 78.0% 54.1% 57.0%45.7% 78.0% 54.1% 57.0%

-100

0

100

200

0 100 200

H e a d

_ A c c Z

Time [ms]

56.9% 60.5% 20.8% 43.2%56.9% 60.5% 20.8% 43.2%


40


47/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

74.0% 71.4% 50.0% 63.5%74.0% 71.4% 50.0% 63.5%

-0.8

-0.4

0

0.4

0 100 200


Time [ms]

83.3% 96.9% 85.8% 87.2%83.3% 96.9% 85.8% 87.2%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

34.5% 80.0% 29.4% 43.2%34.5% 80.0% 29.4% 43.2%

-0.4

-0.2

0

0 100 200

T 1

_ A d i s Y

Time [ms]

88.7% 94.3% 79.2% 86.0%88.7% 94.3% 79.2% 86.0%


-40

0

40

80

0 100 200

H e a d

_ A c c X

Time [ms]

82.3% 64.6% 67.7% 70.5%82.3% 64.6% 67.7% 70.5%

-100

0

100

0 100 200

H e a d

_ A c c Z

Time [ms]

68.2% 61.8% 19.2% 45.2%68.2% 61.8% 19.2% 45.2%


41


48/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

96.3% 98.6% 29.6% 59.3%96.3% 98.6% 29.6% 59.3%

-0.75

-0.5

-0.25

0

0.25

0 100 200


Time [ms]

82.7% 80.0% 85.8% 82.7%82.7% 80.0% 85.8% 82.7%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

34.9% 78.9% 29.6% 43.4%34.9% 78.9% 29.6% 43.4%

-0.3

-0.2

-0.1

0

0.1

0 100 200

T 1

_ A d i s Y

Time [ms]

99.9% 90.2% 79.8% 87.0%99.9% 90.2% 79.8% 87.0%


-40

0

40

80

0 100 200

H e a d

_ A c c X

Time [ms]

77.8% 63.4% 67.4% 68.9%77.8% 63.4% 67.4% 68.9%

-100

0

100

0 100 200

H e a d

_ A c c Z

Time [ms]

58.1% 65.3% 21.3% 44.7%58.1% 65.3% 21.3% 44.7%


42


49/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

98.8% 99.4% 31.6% 60.5%98.8% 99.4% 31.6% 60.5%

-0.75

-0.5

-0.25

0

0.25

0 100 200


Time [ms]

82.9% 80.8% 86.1% 83.1%82.9% 80.8% 86.1% 83.1%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

41.1% 87.3% 33.9% 48.4%41.1% 87.3% 33.9% 48.4%

-0.3

-0.2

-0.1

0

0.1

0 100 200

T 1

_ A d i s Y

Time [ms]

96.1% 90.4% 76.5% 85.2%96.1% 90.4% 76.5% 85.2%


-40

0

40

80

0 100 200

H e a d

_ A c c X

Time [ms]

75.2% 63.3% 67.0% 68.1%75.2% 63.3% 67.0% 68.1%

-100

0

100

0 100 200

H e a d

_ A c c Z

Time [ms]

61.2% 64.3% 21.9% 45.6%61.2% 64.3% 21.9% 45.6%


43


50/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

97.5% 99.6% 32.3% 60.9%97.5% 99.6% 32.3% 60.9%

-0.75

-0.5

-0.25

0

0.25

0 100 200


Time [ms]

82.1% 81.1% 85.4% 82.8%82.1% 81.1% 85.4% 82.8%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

41.8% 87.3% 34.1% 48.7%41.8% 87.3% 34.1% 48.7%

-0.3

-0.2

-0.1

0

0.1

0 100 200

T 1

_ A d i s Y

Time [ms]

96.0% 90.4% 76.6% 85.2%96.0% 90.4% 76.6% 85.2%


-40

0

40

80

0 100 200

H e a d

_ A c c X

Time [ms]

53.5% 73.5% 59.8% 61.3%53.5% 73.5% 59.8% 61.3%

-100

0

100

0 100 200

H e a d

_ A c c Z

Time [ms]

49.0% 66.4% 25.8% 44.5%49.0% 66.4% 25.8% 44.5%


44


51/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

85.5% 85.3% 29.0% 57.3%85.5% 85.3% 29.0% 57.3%

-0.75

-0.5

-0.25

0

0.25

0 100 200


Time [ms]

79.4% 99.0% 86.3% 85.7%79.4% 99.0% 86.3% 85.7%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

85.8% 81.9% 26.8% 55.7%85.8% 81.9% 26.8% 55.7%

-0.4

-0.2

0

0 100 200

T 1

_ A d i s Y

Time [ms]

96.6% 87.4% 89.9% 90.5%96.6% 87.4% 89.9% 90.5%


-40

0

40

80

0 100 200

H e a d

_ A c c X

Time [ms]

49.1% 75.0% 57.6% 59.1%49.1% 75.0% 57.6% 59.1%

-100

0

100

0 100 200

H e a d

_ A c c Z

Time [ms]

40.9% 63.2% 21.4% 39.4%40.9% 63.2% 21.4% 39.4%


45


52/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

98.6% 79.6% 26.8% 56.1%98.6% 79.6% 26.8% 56.1%

-0.75

-0.5

-0.25

0

0.25

0 100 200


Time [ms]

78.9% 99.4% 84.9% 85.0%78.9% 99.4% 84.9% 85.0%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

80.3% 80.8% 21.4% 51.9%80.3% 80.8% 21.4% 51.9%

-0.4

-0.2

0

0 100 200

T 1

_ A d i s Y

Time [ms]

95.1% 97.3% 86.3% 91.5%95.1% 97.3% 86.3% 91.5%


-40

0

40

80

0 100 200

H e a d

_ A c c X

Time [ms]

49.7% 74.8% 57.5% 59.3%49.7% 74.8% 57.5% 59.3%

-100

0

100

0 100 200

H e a d

_ A c c Z

Time [ms]

42.5% 64.4% 22.3% 40.5%42.5% 64.4% 22.3% 40.5%


46


53/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

99.1% 79.5% 26.3% 55.8%99.1% 79.5% 26.3% 55.8%

-0.75

-0.5

-0.25

0

0.25

0 100 200


Time [ms]

79.4% 99.2% 85.1% 85.3%79.4% 99.2% 85.1% 85.3%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

82.3% 81.8% 22.0% 52.6%82.3% 81.8% 22.0% 52.6%

-0.4

-0.2

0

0 100 200

T 1

_ A d i s Y

Time [ms]

94.5% 97.2% 85.6% 90.9%94.5% 97.2% 85.6% 90.9%


-40

0

40

80

0 100 200

H e a d

_ A c c X

Time [ms]

56.8% 97.7% 55.1% 64.0%56.8% 97.7% 55.1% 64.0%

-100

0

100

0 100 200

H e a d

_ A c c Z

Time [ms]

80.4% 79.7% 26.8% 54.7%80.4% 79.7% 26.8% 54.7%


47


54/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

88.8% 97.2% 44.4% 67.2%88.8% 97.2% 44.4% 67.2%

-0.75

-0.5

-0.25

0

0.25

0 100 200


Time [ms]

76.5% 100.0% 81.0% 82.6%76.5% 100.0% 81.0% 82.6%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

87.7% 83.3% 36.7% 61.6%87.7% 83.3% 36.7% 61.6%

-0.4

-0.2

0

0 100 200

T 1

_ A d i s Y

Time [ms]

97.0% 90.5% 85.9% 90.0%97.0% 90.5% 85.9% 90.0%


-40

0

40

80

0 100 200

H e a d

_ A c c X

Time [ms]

45.0% 99.4% 53.0% 58.3%45.0% 99.4% 53.0% 58.3%

-100

0

100

0 100 200

H e a d

_ A c c Z

Time [ms]

99.1% 79.0% 22.9% 53.9%99.1% 79.0% 22.9% 53.9%


48


55/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

86.3% 98.1% 42.4% 65.8%86.3% 98.1% 42.4% 65.8%

-0.75

-0.5

-0.25

0

0.25

0 100 200


Time [ms]

78.8% 83.5% 83.2% 81.7%78.8% 83.5% 83.2% 81.7%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

86.7% 84.4% 34.3% 60.3%86.7% 84.4% 34.3% 60.3%

-0.4

-0.2

0

0 100 200

T 1

_ A d i s Y

Time [ms]

94.4% 90.8% 85.1% 89.4%94.4% 90.8% 85.1% 89.4%


-40

0

40

80

0 100 200

H e a d

_ A c c X

Time [ms]

50.7% 84.4% 55.0% 60.4%50.7% 84.4% 55.0% 60.4%

-100

0

100

0 100 200

H e a d

_ A c c Z

Time [ms]

99.0% 79.9% 25.0% 55.2%99.0% 79.9% 25.0% 55.2%


49


56/87


-0.5

-0.25

0

0.25

0.5

0 100 200

H e a d

_ A d i s Y

Time [ms]

92.6% 99.6% 42.6% 66.6%92.6% 99.6% 42.6% 66.6%

-0.75

-0.5

-0.25

0

0.25

0 100 200


Time [ms]

77.5% 98.8% 82.9% 83.6%77.5% 98.8% 82.9% 83.6%


-100

0

100

200

300

0 100 200

T 1

_ A c c X

Time [ms]

89.0% 85.2% 35.8% 61.5%89.0% 85.2% 35.8% 61.5%

-0.4

-0.2

0

0 100 200

T 1

_ A d i s Y

Time [ms]

94.9% 90.0% 86.5% 89.9%94.9% 90.0% 86.5% 89.9%


-40

0

40

80

0 100 200

H e a d

_ A c c X

Time [ms]

66.1% 57.0% 51.0% 57.6%55.8% 81.2% 55.5% 62.2%

-100

0

100

Biorid II Qr

Documents

Transcript of Biorid II Qr