Download this VSM example:   Value Stream Map and Modeling

Value Stream Maps are typically designed to flow horizontally left to right. When a process is complex it becomes difficult to view the whole process: it may require several pages and this makes it difficult to see the total picture.

Making the VSM vertically allows us to visualize complex processes in a single page.

Another common practice is to map the process by sticking post-its on a wall or large wrapping paper. This has the advantage of simplicity but it creates some inconveniences. It becomes impractical to save all the work done to continue working on it later on. It is also difficult to communicate to others: photos are difficult to read. If you ask other people, not present in the VSM exercise, to give their input they can not modify the work you have done.

Doing a VSM in Excel has several advantages:

  1. The team involved in the VSM exercise can be located in remote locations and can all share a spread sheet in the cloud with communication via Hangouts or Skype.
  2. All the work done is not lost at the end of the meeting and it can be updated remotely at a later stage by all process participants
  3. Excel allows the addition of the process parameters for each process step by adding as many columns as required on the right:

·         Process time

·         Setup time

·         Cycle time

·         Work-In-Process waiting

Complex processes require often feedback loops: repair failing items and test them again. These loops are an essential part of the VSM because they often become the bottleneck of the total process. 

When the process doesn’t have a single flow but there are several branches you need to estimate the proportion of items that will flow along each branch and this you can do by collecting data along a period of time.

Process times have variability so you will need to estimate not only their average values but also their standard deviations.

 

In this Repair process the average time it takes to receive a product is 2 min and its standard deviation 0.5 min.

20% of the products are under warranty so 80% will follow the “N” branch.

Average inter arrival time of products coming from the customer for repair is 5 min. This is the takt time the process has to comply with. 

 

Not all products go through all process steps, therefore we need to calculate % throughput for each step based on the % of products flowing along each branch.

In order to balance the line we need the right staffing on each process step to insure enough capacity to handle the product arrivals. This means that Cycle time  Inter arrival time (takt).

Cycle time  =  Average time   x   % Throughput  /  Staffing 

We can calculate the staffing (Full Time Equivalent) required for each step M4 – M13 with Solver

 To do this we calculate the total staff required in cell M1 just adding all the yellow cells in column M. This cell is what we want to minimize as long as we meet the requirements that all cycle times P4:P13  are below the inter arrival time in K3. 

 We are adding the constraint M4:M13 0.1. The proposed staffing by Solver in the yellow cells M4:M13 is:

Which gives a total staff of 13.6 (M1). This is a theoretical minimum which assumes all staff has all skills, which is not the case.

In cells D1 to H1 we obtain the compound for each process participant (some perform several operations) obtained from the column M.

If Receive and Coordinator can be combined we may have to round up the sum of the two from the 1.6 requirement to 2.0

Invoicing we round up from 1.8 to 2. Test from 2.9 to 3 and repair from 7.4 to 8. 

This manual rounding will lead us to this situation:

Where the total staff has increased from the theoretical 13.6 to 15 after considering the different skills required in the different steps.

This increase in the staffing has an effect on lowering the cycle of the steps affected below the takt time of 5 minutes.

Looking at the % Throughput column we notice that some steps only process 4% of the products to be repaired while test and repair have to process 123%. The reason for this is that only 80% pass the test OK the first time. This means 20% have to be tested a second time. Same with burn-in, where 10% fail and have to be retested.

 So far we have not taken into account variability but we will do so by adding some additional columns to our VSM converting it into a simulator in our next article: How to convert your VSM into a simulator to experience variation