Reliability measurement at virtual phase

Anitha.A, Nagendra Pansare, Vivek Gupta

Mahindra and Mahindra Ltd, MRV

ABSTRACT

Reliability is a significant attribute for Agricultural Tractor since it uses its own generated traction force for different kinds of implements to perform as per customer required functions. Once the Reliability target is done, the DFMEA, virtual lab analyses of all possible failure modes and design calculations are the important source of information for virtual reliability measurement. The purpose is to produce error free robust design and to identify the required physical test. A statistical model has been developed to estimate virtual reliability of the product as well as virtual lab through hit ratio and coverage ratio. This model based measurement helps to track the reliability level at virtual phase and the focus area of the virtual lab. This analysis also assists the manufacturer to save development time and cost by reducing the number of required prototype tests as well as field test during development phase of product life cycle.

  1. INTRODUCTION

In an organization, the reliability is tracked in terms of Reliability number or Repair per Hundred(RPH) (in case of agricultural tractors industry). Once the reliability goal is set and allocated for each systems through reliability parenting process, then tracking reliability plays a major role for the manufacturer business to run and extend the future. The Design Failure Mode Effect Analysis (DFMEA) of all systems are ready once the design is confirmed for the product. In a product life cycle, with an advanced technology, the products are going through several software simulations and design calculations before proto type build. These virtual simulations are matured by the accuracy of virtual software also it relies on the capability of virtual software, capacity of user and the capacity of user to use the virtual software. This enables the manufacturer to track the efficacy of virtual software also to measure the reliability at virtual phase.

  1. BACKGROUND

All physical system are subjected to different series of stress in its life cycle when it is in operable condition. The tractor system is generally subjected to stresses like thermal, vibration due to different kinds of its load and applications. Eventually failures occurs due to subjected stresses and it is based on the resistance of a structure of a system versus the load applied to the system. Since the strength and load are typically functions of time, the probability of failure can be estimated through software simulations virtually and at the same time the reliability is concerned with how long the product continues to function once it becomes operational. During new product development, it is important to know the reliability at virtual phase since it covers 100 percent design failure modes. Hence to validate the design failure modes which includes all structural components failures due to fatigue, mounting and servicing issues, it is required to find the reliability level at virtual phase against the target. The design failure modes that are related to system structure and performance, its assembling and servicing are captured through DFMEA, Real World Usage Pattern (RWUP) and it is validated through virtual simulations (CAE, NVH, CFD, Dynamics, etc.) output and design calculations which are the key sources of virtual reliability.

  1. Virtual Reliability PROCESS

3.1Virtual reliability process flow

Figure.1. Virtual reliability process flow

The Figure.1 shows the process flow of doing virtual reliability analysis during new product development

The method of estimating virtual reliability for a new product as follows:

(1)Virtual reliability analysis begins with setting reliability target for the new product through reliability parenting process

  • Reliability parenting process is the collective information analysis used to have accurate Reliability target setting which is based on Customer expectations, Competition performance, Technology used, Cost constraints, Usage patterns and the RPH of Parent model.

(2)Once the concept design is done, then conduct DFMEA for all aggregates and subsystems. Then the failure modes are segregated associated with their aggregates and entered in to the Virtual reliability template for the virtual lab inputs

(3)Identify the scope of virtual labs (CAE, NVH, CFD, Materials Lab and Dynamics) based on DFMEA. This will provide the number of the applicable failure modes which means the capability of doing virtual validation from each virtual lab.

(4)Analyze the targeted failure modes from the applicable failure modes by Virtual labs and monitor the actions on failure modes.

(5)Find the success probability from virtual labs clearance on the aimed failure modes

(6)Estimate Virtual reliability of tractor systems and for eachvirtual lab.

(7)Finally the results are compared with targeted reliability and helps the way to enhance the virtual reliability at virtual phase by improving the power of virtual lab and virtual lab capability with respective to virtual labs.

(8)The power of virtual lab gives the effectiveness of virtual validation by verifying the failure modes physically.

(9)This also enables the virtual lab to identify future competency and capability planning

3.2 Reliability target setting

Reliability target for new product (Tractor) is divided into systems and subsystems. In tractor, the engine, transmission, hydraulics, electrical, Sheet metal and overall tractor are considered as systems and each of their subsystems has its own identical components. It is assumed that all systems and subsystems are connected in series as a worst case for analysis.

The stable RPH of the parent model is evaluated from the matured warranty data. The target RPH for new product is derived by considering all design changes from parent model, Real World Usage Pattern(RWUP), Annual Average Usage Hours(AAHU), cost and maintenance. The RPH criteria for the new product is developed as in Table 1 below.

.

Rating / Criteria
Much better than the parent (MB) / 70% less than the parent RPH
Better than the parent (B) / 30% less than the parent RPH
Same as the parent (S) / Same RPH as that of parent
Par with competitors(Tradeoff) (T) / Same RPH as competitor

Table 1. RPH Criteria for new product

Eventually the reliability target is set as RPH value for each subsystem based on RPH criteria through reliability parenting process.

3.3Virtual reliability

Figure 2.Pictorial representation of Virtual reliability scope

Finding virtual reliability is an interesting scope against the product reliability target. In the above figure depicts four zones which are described as below:

  • Z1:Zone 1. This represents the number of failure modes that are cleared by virtual lab.
  • Z 2: Zone 2. The Zone 1 plus Zone 2 represents the targeted number of failure modes by virtual lab
  • Z 3: Zone3. Now Zone 1, Zone 2 and Zone 3 together represents the applicable number of failure modes by virtual lab
  • Z 4: Zone 4. This is an open area where the number failure modes can only be validated through either field test or lab test.

Virtual reliability is estimated by taking the product of virtual lab capability proportionate and success probability as below.

Virtual Reliability = Virtual lab capability proportionate * Success probability

Where,

Virtual lab capability proportionate: It is a ratio of targeted number of failure modes from number of applicable failure modes which are in line with DFMEA. This shows the confidenceon each failure mode by virtual Lab.

Success probability: It is the ratio of cleared number of failure modes by virtual lab from its targeted number of failure modes. Hence it provides the success probability at virtual phase. Here the probability value can only take either 0 or 1 for each failure modes since the virtual lab can have either pass or fail as outcome.

Power of virtual Lab: It is a ratio of successful failure modes which are verified through either lab or field to the total number of cleared failure modes by virtual lab.

The power of virtual test is been taken as either zero or one for each failure mode.

Hence virtual reliability is defined as the product of virtual lab capability proportionateand success probability at the specified duration by the management in the organization. The duration given for virtual reliability varies in each organization.

  1. Case study

4.1Reliability targets setting

The reliability targets are done for one of the project through Reliability parenting process based on RPH criteria as Table 2 below

S.No / Tractor systems / Subsystems / Parent model - Stable RPH / New Product - Target RPH / % Improvement over parent model / Target Reliability
1 / Engine / Engine / 12.14 / 12.14 / 12.14 / 12.14 / No change done / 89%
2 / Transmission / Rear Axle / 0.4 / 17.39 / 0.4 / 12.46 / 28% / 88%
Gear Box / 15.5 / 10.85
Differential / 0.93 / 0.651
Clutch / 0.56 / 0.56
3 / Hydraulic / Lift Unit / 12.2 / 14.88 / 3.66 / 5.54 / 63% / 95%
Hydraulic linkages / 2.68 / 1.876
4 / Overall Tractor / Steering / 0.38 / 1.21 / 0.38 / 0.96 / 21% / 99%
Front Axle / 0.83 / 0.581
5 / Sheet Metal / Fender / 1.35 / 1.35 / 1.35 / 1.35 / No change done / 99%
6 / Electricals / Electricals / 4.5 / 4.5 / 3.15 / 3.15 / 30% / 97%
Tractor level / 51.47 / 35.598 / 31% / 70%

Table 2. Reliability target setup

In the above table 2, it show the reliability target for the new product and the % improvement over the parent model for each subsystems of Tractor through reliability parenting process.

4.2Finding scope of Virtual lab

Firstly the failure modes from DFMEA of each aggregates are entered in to the virtual reliability template. Then all other inputs are entered by virtual labs which are specific to the project.

S.No / System / Sub-system / Failure Mode from DFMEA / Is it possible to validate virtually? / CAE / NVH / CFD / Dynamics / No of attempt / No. of targeting failure modes
1 / Tractor Overall / Front-Axle / Steering effort more than 5 kg. / Y / 1 / 1
2 / Front axle unable to withstand 450 kg load / Y / 1 / 1 / 2 / 1
3 / Swivel angle less than 8 deg. / Y / 1 / 1 / 1
4 / Front axle fails before desired (6000 hrs) life / Y / 1 / 1 / 1
5 / Steered wheel not operational / N / 0 / 0
6 / Steered wheel partially operational / N / 0 / 0
7 / Front axle fails to withstand Shock load at the time of front end lifting & declutching / Y / 1 / 1 / 1
8 / Turning radius more than 2.6 m / Y / 0 / 1

Table 3. Virtual reliability template

In Table 3, the column “Is it possible to validate virtually” is providing the virtual labs capability to validate the failure modes and this together with total number of targeting failure modes is used to derive virtual lab capability proportionate and the success probability is found by the ratio of successful failure modes from targeted number of failure modes.

4.3 Finding virtual targets

In the specific project the virtual reliability scope has been found out from the observation below:

The total available failure modes in DFMEA = 565

The total applicable failure modes in virtual = 403

  • This gives the Virtual lab validation capability as 71%. In other words it helps to make reliability target at virtual phase as 50% against target. Hence the remaining 29% will be validated by Field and Lab only.

From total applicable failure modes, Virtual lab is targeting 374 failure modes.

  • This provides the Virtual lab capability proportionate as 374/403 = 91%

From targeted number of failure modes, Virtual lab is clearing 362 failure modes

  • This provides the success probability as 362/374 = 97%

From the cleared failure modes,20 failures are happening in either lab or field from 362 failure modes

  • This gives the power of virtual lab as 342/362 = 95%

5Results and discussion

5.1 Virtual reliability of Tractor systems

Target for Probability of Success for all systems are 100%. Expected virtual reliability of each tractor system is 90%. Hence the systems below 90% virtual reliability needs to have necessary action to clear the failure modes.

Tractor Systems / Total failure modes in DFMEA / Applicable no. of failure modes / No. of targeted Failure modes / Virtual Capability proportionate / No. of cleared failure modes / Prob. of Success / Virtual reliability
Transmission / 87 / 80 / 75 / 94% / 74 / 98.67% / 92.50%
Hydraulics / 98 / 67 / 57 / 85% / 57 / 100.00% / 85.07%
Engine / 108 / 82 / 80 / 98% / 78 / 97.50% / 95.12%
Tractor overall / 84 / 74 / 69 / 93% / 66 / 95.65% / 89.19%
Sheet metal / 110 / 80 / 75 / 94% / 70 / 93.33% / 87.50%
Electrical / 78 / 20 / 18 / 90% / 17.00 / 94.44% / 85.00%

Table 4. Virtual reliability of all systems in Tractor

In table 4. the success probability is not meeting the requirement for all systems except Hydraulics system, however the transmission system meeting the expected virtual reliability. Hence all other systems need to have either improvement in the design action or improvement in lab by clearing the failure modes through physical testing.

Reliability value
Target reliability of Tractor / 70.00%
Virtual target reliability / 50%
Achieved virtual reliability / 46%

Table 5. Virtual reliability against target reliability

The virtual reliability results as 46% achieved which indicates as the product design is as good as reliable and it identifies the improvement area of tractor system.

5.2 Virtual reliability of the virtual labs

The table shows the result of virtual labs

Virtual Lab / AchievedVirtual reliability
CAE / 73%
Material / 80%
NVH / 60%
Dynamics / 90%
CFD / 100%

Table 6.Virtual reliability of virtual lab

From table 6. it is apparent that virtual reliability requires certain improvement for CAE, Material lab and NVH. In this case to improve virtual reliability it is required to relook the power of virtual lab and the virtual lab capability.

Figure 3. Virtual reliability of virtual labs

Target virtual reliability for the virtual lab is 100% and the expected is 90%. The above fgure 3 shows that the Dynamics and CFD are meeting the requirements. CAE, NVH and Materials lab not meeting the requirments and are requested to check theefficacy of design software that are mapping with RWUP, Power of virtual lab and the effectiveness of validating applicable failure modes that are inline with DFMEA.

Conclusion:

In order to ensure time to market and have robust design, it is recommended to improve virtual lab capabilityand the power of virtual lab by validating maximum number of failure modes that are linedup with DFMEA and . Having virtual reliability estimation provides the direct focus on necessary and improvement area of doing reliability test.This analysis helps to have more reliable design and avoid unnecessary expenses due to rework on design change and the duration of test. Also this methodology is used to have long term vision for the manufacturer and to find the efficacy of design software which are aligned with the knowledge of RWUP.

References

  1. Modal analysis of agricultural machineries using finite element method: A case study for a V-belt pulley of a fodder crushing machine, Journal of Food, Agriculture & Environment Vol.8 (3& 4 ) : 4 3 9 - 4 4 6 . 2 0 1 0
  2. A Case Study of Reliability Analysis on the Damage State of Existing Concrete Viaduct Structure, Tamkang Journal of Science and Engineering, Vol. 12, No. 4, pp. 371_379 (2009)
  3. A Novel Virtual Age Reliability Model for Time-to-Failure Prediction,Yao Wang, Sorin Cotofana, Computer Engineering Laboratory, EEMCS Delft University of Technology

CONTACT

Anitha.A is a Lead Engineer in Reliability, Vehicle Integration Department, Farm sector, Mahindra & Mahindra Ltd., Chennai, Tamilnadu, India. She received her M.Tech. Degree in Reliability Engineering from Indian Institute of Technology, Kharagpur. Her responsible areas at Mahindra and Mahindra is to establish Reliability process through Reliability target setting, Virtual reliability estimation, Reliability growth analysis, Reliability prediction, Warranty data analysis & Weibull analysis.

Contact:

Abbreviations

DFMEA: Design Failure Mode Effect Analysis

PRH: Repair Per Hundred

RWUP:Real World Usage Pattern

CAE: Computer Aided Engineering

NVH:Noise, vibration, and harshness

CFD: Computational fluid dynamics