Examples for Reliability Growth

The following example demonstrates a scenario where you would create a Reliability Growth Analysis with event-based data that is measured using failure dates. 

If you track events (e.g., safety events or failures) by specific date, then you can create a Reliability Growth Analysis using event-based data that is measured using failure dates.

Centrifugal Pump 1051700 was installed at the Springfield plant on July 20, 1998. Since its installation, the pump has failed several times, and each time the pump fails it is repaired immediately without any significant downtime. You have collected data noting which days the pump failed. You want to use the following data to perform a Reliability Growth Analysis to determine the mean time between pump failures.

Failure Dates	Number of Failures
12/14/1998	1
2/7/1999	1
6/26/1999	1
8/1/1999	1
9/5/1999	1
1/1/2001	1
1/2/2001	1
2/7/2002	1
7/11/2002	1
12/10/2002	1
5/12/2003	1
7/2/2003	1
11/28/2005	1
1/30/2006	1

In this case, you would use event-based data derived from failure dates to predict:

• The date when the pump will fail next.
• How many times the pump will fail by a certain date.
• If the number of failures increased or decreased as a result of a strategy change (e.g., a new preventative maintenance schedule), which was implemented after the failure on September 5, 1999.

The following example demonstrates a scenario where you would create a Reliability Growth Analysis with event-based data that is measured using cumulative operating time.

Sometimes the specific days on which a piece of equipment or location fails may not be available, or the piece of equipment or location may not operate for the same amount of time every day. In these cases, it would be more valuable to predict future failures using cumulative operating time (COT), or the amount of time the piece of equipment or location has been in operation.

Haul Truck 1 was purchased for a shipping firm as a used vehicle with 11,028 miles. The truck now runs varied routes depending on the number of shipments to which it is assigned. Because the truck does not always travel the same number of miles each day, you collect the data representing the mileage points at which the truck broke down. You want to use the following data to perform a Reliability Growth Analysis to determine the mean operating time (i.e., number of miles) between failures.

Cumulative Operating Time (Miles)	Number of Failures
27,393	1
41,675	1
60,400	1
66,128	1
72,132	1
77,314	1
113,888	1
146,671	1
205,424	1
224,624	1
243,841	1
260,828	1
279,951	1
303,156	1

In this case, you would use event-based data derived from cumulative operating time to predict:

• The mileage point at which the next failure will occur.
• How many failures will occur by a certain mileage point (e.g., five failures will have occurred after 1,000 miles).
• If the number of failures increased or decreased as a result of a strategy change (e.g., more routine oil changes), which was implemented after the failure at 77,314 miles.

In Examples 1 and 2, each datapoint represents a single measurement or failure. In some datasets, each datapoint may represent more than one measurement, or an amount of data. Throughout this documentation, this type of data is referred to as grouped data. To perform Reliability Growth Analyses on grouped data, when you create a dataset, you must use datapoints that represent multiple measurements or an amount of data. Datasets containing grouped data can be based on either failure dates or cumulative operating time.

For example, if you want to analyze a pump for which you record data every six months and every time you record data you record multiple failures (one datapoint represents multiple failures), you would create a Reliability Growth Analysis based on event-based, grouped data that is derived from cumulative operating time.

In Examples 1 and 2, each datapoint was also based on an event. Examples of event-based failures include equipment or location failures, safety incidents, or equipment or location repairs. Sometimes, you may want to perform a Reliability Growth Analysis on a variable that does not measure a specific event (e.g., an amount). For the correct labels to appear throughout the analysis, these datasets should be entered as non-event data. Datasets containing non-event data can be based on either failure dates or cumulative operating time.

Using a Reliability Growth Analysis to measure cost is the most common example of evaluating grouped data and non-event data.

The following example demonstrates a scenario where you would create a Reliability Growth Analysis with grouped data that is not event-based that is measured using cumulative operating time.

You want to measure the cost of equipment failures at the Springfield plant in order to determine how you should plan a budget for that plant in the future. Individual pieces of equipment fail at different rates, so you measure the total cost of replacement parts and mechanic labor for the entire plant every few months. You want to use the following data to perform a Reliability Growth Analysis to view the cost trends at this plant.

Cumulative Operating Time (Months)	Cost
6.5	1,120
13	996
17	1,052
23	1,085
37	1,075
49.5	1,096
62	1,001

In this case, you would use non-event, grouped data that is derived from cumulative operating time to predict:

• The repair costs that the plant will incur by some point in the future.
• If the repair costs have increased or decreased as a result of a strategy change (e.g., changing the mechanic who replaces parts), which was implemented after 23 months.