Utilizing Process Behavior Chart to Forecast Weekly Shipments & Set Safety Stock

[Danoend@charter.net]

I have been reading the DEN message archive for about 7
 years.  I would like some feedback on an approach I have
 been using in my work as a production scheduler/planner.


I have been entering a forecast for distribution items
 equal to the Center Line of a process behavior chart.  The
 data includes weekly shipments from January 2002 through
 the present.  Weekly shipments of most of the items seem
 stable, only a point or two of the data set above the
 upper control limits and no runs of eight points above or
 below the center line.  Investigating the points above the
 upper control limit usually reveals 2 or 3 exceptionally large qty customer orders in the same week.


I have been scheduling production to satisfy the average
 weekly shipments and have therefore set the safety stock
 using the following equation: Safety Stock = UNPL - CL.
  So we are producing what we expect to ship each week, and
 covering the amount that the shipments could be expected
 to vary to with safety stock.


My materials manager liked the approach because the method
 made sense to him and reduced the $ value of safety
 stocked finished goods by $47,000.  It theoretically
 provided better protection against stock outs.  Some items
 had safety stock increase, some decreased.  I have been
 creating the process control charts in excel.  The
 majority of our sales volume comes from 233 items.  The thing my boss did not like is the
 amount of time it takes me to create the charts in excel for this many items.



We have a new Information technology manager, who has taken
 on the task of adding sophistication to the forecasting and
 safety stock calculation.  I warned him that if he used a
 global measure of dispersion that out of control points
 would inappropriately inflate his calculation of the 3
 sigma value.  I do not think he understood what I was
 telling him.  He has provided my boss with a time series
 chart of weekly shipments for a particular item, with trend
 lines at various points along the chart (not sure what
 criteria he used to decide where to put these lines) and a
 suggested safety stock level based on different service
 levels (calculation ?).  The process behavior chart for
 this particular item does not indicate any trends.


Is the method of using a process behavior chart as laid out
 above to set the forecast and safety stock make sense?



I suspect the new system, seeing trends where they are not
 indicated on the process behavior chart, and frequent
 adjusting of the forecast and safety stock levels is
 tampering with a stable system.  Could this be so?


I am trying to implement what I have learned reading
 Shewhart/Deming/Wheeler into my work.  I am convinced of
 the value of the process behavior chart approach.  How do
 I convince my boss that the higher perceived sophistication of the proposed new method is inferior to
 the process behavior chart approach?



Can anyone recommend a book that integrates the process
 behavior chart/analytical statistical approach with
 materials management/production planning & control
 processes?


I am grateful for any feedback on these issues.

Thank you,
Dan Enderle
Danoend@charter.net


Message posting through the Clemson CQI Web Server.