CHAPTERS IMPLEMENTATION
CHAPTER 7. DEMONSTRATION: A CHARGING AND STARTING SYSTEM
7.5 The Masking Process
Guess et al. [1991] state that it is important to understand the mechanism causing masking to occur. It can be used to verify that assumptions concerning masking probabilities are valid and that estimators are consistent. A few comments can be made with regards to the masking process observed in this case study. The warranty claim process that leads to masking is shown in Figure 7.2.
The truck manufacturer is only aware o f those failure events that lead to a warranty claim. The warranty period on this particular tractor is three years. Some aspects o f the repair and warranty claim process can lead to masking o f the cause o f failure for some observations. They are discussed next.
1. Operation
A failure occurs during operation and is noticed by the driver. He’ll usually visit the dealership where the vehicle was purchased or visits a service center to have it corrected. Note that the failure might be caused by incorrect operation o f the vehicle. Some drivers install a large number o f electronic accessories, which might affect the reliability o f the Charging and Starting System. A driver might misrepresent the failure circumstances to the diagnostician to hide an operator error. This complicates the fault-finding process.
155
Repaired Truck Stressed Components Problem Noticed Problem Described R ep air / O rd er / s ' W arranty Claim creati create update dose/consult Cause Identified update/dose R ep air or R ep lacem en t Completed _ Repair Work Submitted Claim dose —Paid Claim* Completed Repair Work
Incorrectly Paid Claim
-Unretumed ParV
Analyzed Part Replaced Part Returned
Failure A nalysis > R eport / ■create C ost R ecovery Item R eturned to V endor D eterm ine Responsibility S ervice D iagnostics Item S c ra p p e d Claim P ay m en t Claim S ubm ission Failure E vent Visit D ealership o r F leet S ervice Shop Failure Analysis Truck O peration
Figure 7.2: Service Process Leading to Masked Data
156
2. Diagnosis
The truck is diagnosed, possibly with the help o f an expert system. The diagnostic equipment that is available at a service center or dealership, is sometimes limited in capability. The failure might be hidden and difficult to identify. In addition, the person performing the analysis m ight have limited interest or experience in diagnosis. All the while, the customer, who is eager to get back on the road, is expecting a diagnostic result that identifies a faulty component. I f service is done on a truck that’s still under warranty coverage, a dealer can only charge the manufacturer for (diagnostic) work performed. He cannot charge the driver. These factors put pressure on the diagnostician to state that a certain part has failed, though diagnostic information might not be conclusive.
3. Claim Description
Initially, a driver complaint on the truck’s performance is registered. Often, a person other than the employee o f the dealership who performed the diagnosis and repair work, enters a claim description into the computer system and submits it for refund. As a result, claim descriptions tend to be vague and brief. The fault code associated with repair activities determines to a large extent which organization (manufacturer, vendor, dealer or client) will be responsible for the costs o f the repair work. The dealership or fleet service shop enters this information. It receives payment by the truck manufacturer in advance o f a possible failure analysis o f the removed component.
4. Returned Parts
Some defective items get returned to the vendor that supplied them. It depends on the cost o f the part, the shipping costs and vendor policy, whether an item is returned or not. In some cases, the wrong part gets returned to the vendor. In other cases, the part gets lost
157
and is never received by the vendor. The person who returns a change-out part is not the same individual who replaced the part.
5. Failure Analysis
Failure analysis is performed on a sample o f the failed items only, even on the more expensive items, due to time and financial constraints. It is difficult to obtain information on the operational environment a major component might have been exposed to. It is even more difficult to test the part under simulated operational conditions. As a result, no fault might be found during failure analysis although one actually exists (i.e. Type II error, see [Iman and Conover, 1983]). In addition, vendors might be motivated to delay (reporting of) failure analysis as long as possible if they suspect the analysis would lead to a cost to them (i.e. so-called ‘Vendor recovery”).
6. Assigning Responsibility for Costs
Initially, the manufacturer pays the dealership or service center for all approved warranty claims. If subsequent failure analysis shows that the vendor was responsible for the costs o f a warranty claim, a so-called vendor recovery action is initiated to get the money back from the vendor. If it shows that a non-defective part was replaced, then the dealer must return the funds paid for the claim.
7. Warranty Data Analysis
Failure diagnostic and warranty claim data are stored in different databases. They are not easily combined or compared. Furthermore, the workload o f reliability engineers prohibits a detailed comparison o f data entries relating to the same failure event. The complexity o f the components, their multitude of failure modes and the shortage of
158
experts with access to all stored failure data, prohibit a detailed failure cause analysis based on stored data.