Process Reliability

The Process Reliability module allows users to identify systematic and reliability losses by recognizing trends in the production data.

The daily production data is analyzed by means of Weibull analysis plots. Nameplate production ratings may be specified (similar to concepts originating from six-sigma methodology) and compared with demonstrated production data. Losses can therefore be identified for corrective action.

The Process Reliability module allows production data to be compared from different plants or different time periods. Production data may also be transferred from simulated production profiles created from within the AvSim module. This allows production data from potential plant improvements to be compared with existing production data.

The Process Reliability module of Availability Workbench analyzes daily plant production data to distinguish between systematic production losses and reliability losses. The method allows users to quickly analyze whether a production process is efficient and distinguish reliability losses from production losses.

Daily production data is analyzed using Weibull and other types of plots. The Cumulative Probability plot is used to identify trends in the data. Cusps in the plot are marked to identify regions (of daily product capacity ranges). The right-most cusp on the Cumulative Probability plot identifies the point at which reliability problems begin to occur (lower daily capacity values to the left of this cusp will be influenced more strongly by reliability problems rather than controlled process losses). The points to the right of the cusp will normally form a reasonably straight line and are fitted (automatically by the program once the cusp has been defined by the user) to a Weibull distribution. These points represent production losses for a controlled process. A steep line is preferred as it indicates a very tight process with little variation in production from day to day.

Users may define a theoretical Nameplate Weibull distribution which is also displayed on the Cumulative Probability plot. The ‘nameplate’ represents the desired or maximum achievable production capacity from a controlled process with no reliability losses. It is not based on the actual plant data but is defined by the user.

In some cases defining a single cusp to mark the onset of reliability losses is sufficient. All data values to the left of the cusp (lower daily production capacities) are assumed to have a reliability loss element. However, in many cases, the reliability loss region may display one or more additional cusps in the Cumulative Probability plot indicating that there are distinct reliability problems in the process. Users may mark such cusps resulting in the program splitting the reliability loss region into two or more regions.

When cusps are defined on the Cumulative Probability plot, Availability Workbench calculates various parameters associated with the entire Process Set (all the daily capacity losses recorded in the set). Principal parameters are Actual Losses, Production Losses and Reliability Losses (all expressed as an average loss per day). Production Losses are determined from the Weibull distribution fitted to the data points to the right of the right-most cusp. Reliability Losses are determined by subtracting Production Losses from Actual Losses. Production Losses are extrapolated into the reliability loss regions to the left of the right-most cusp on the basis that there will be some contribution to actual losses from a controlled production process even though the reliability losses become more dominant when daily losses are more severe. Actual Losses, Production Losses and Reliability Losses are also calculated for each reliability loss region defined to the left of the right-most cusp. Reliability Losses are determined by subtracting Production Losses from Actual Losses in each region.

Screen Shots