Process Flow Models - Downtime

There will be times when a process experiences downtime, that is, time when the resources associated with production becomes idle. Downtime may be caused by a failure of a machine that causes production to stop until the problem is located and repaired. It may also result if some raw material is unavailable. Then production must stop until the raw material is obtained. Usually, downtime is disruptive and should be eliminated if possible, but a correct model should recognize the existence of downtime.

We accommodate downtime by providing a column on the data form. A number entered in the column for a particular operation represents the expected downtime per unit produced. The time is accumulated in the Adjusted Time column and ultimately is included in the Unit Time. Downtime expends the time available on a resource and thus reduces the time available for useful production.

There are mechanisms that cause an entire process to go down. Such times should be included for all operations in the process affected. All downtimes must be expressed as expected times per unit of production.

 To include downtime in the analysis, click the Downtime button on the dialog. To keep the example simple, we do not select the remaining options. The example has five operations arranged in series. The demand for the product is 40 per week and the operating interval is 40 hours per week. With no losses, this results in a flow rate of 1 per hour as shown in column M. The operating times are shown in column F. If each operation were implemented in a separate station, the WIP in column N shows that all the stations average 1 or less product, so one station for each operation is sufficient when all downtimes are equal to 0.

For purposes of illustration, assume that 1% of the items passing through the line will cause an event that results in the entire line being shut down. Further assume that the average downtime caused by this event is 2 hours. With a flow of 40 per week, the expected downtime for the system is 40*0.01*2 = 0.8 hours per week. Expressed as a per unit amount this is 0.8/40 = 0.02 hour per unit.

A second event that occurs at a rate of 1% of the units process, causes only operation 4 to shut down. This downtime is 10 hours for each event. By the same reasoning this adds an additional 0.01*10 = 0.1 hours per unit at operation 5. The resulting downtimes are added on the form below.

The WIP values now indicate that a single station is no longer sufficient for operations 1 and 4 since the values are greater than 1 for these operations.

Not shown by this analysis is the disruptive effect of the randomness of the downtime events. When the entire process shuts down, the materials in process remain at the idle production stations, so when the process restarts there is no marked redistribution of WIP. When a single station is affected and the other stations continue processing, problems of supply arise when stations downstream of the downed station become starved and inventory piles up at the downed station. This analysis does not show this effect.

When disruptive events concerning a single station are allowed, inventory must be maintained to buffer the effects of downtime. The add-in does provide an inventory option through the Time Function feature, but the parameters of the inventory must be externally set.

Operations Management / Industrial Engineering
Internet
by Paul A. Jensen