4a. Flow-Rate Acceleration

Here’s a little more detail on the Flow-Rate Acceleration Method (minor refinements on the Goldratt/TOC method).

The Concept: A reminder

Flow-Rate Acceleration is about improving the number of items/jobs processed per time by the system – the throughput of the system.

The flow-rate of any system is limited by the flow-rate through the bottleneck.

The only way to improve the end-to-end flow-rate of the system is to improve the flow-rate at the bottleneck.  Improving it anywhere else will have no positive effect on flow – and will probably have a negative effect.

The example below is for a simple linear flow, but the method is identical for complex flows.

The Method

1. The first step is to draw the flow-system.  Note that the blocks in the diagram are resources – machines or people – in the system.

2. The next steps is to find the bottleneck.  The numbers in the blocks are the number of units that that resource can produce per hour or day.  If it’s not immediately obvious where the bottleneck is, look for the biggest backlog or find the resource that has the lowest unit/per time rate.  In this case, it’s resource no 4 – which can only do 2 units per hour.

3. The third step is to introduce Quality Control immediately before the bottleneck – to ensure that precious bottleneck capacity isn’t wasted processing items that are already destined for rework or the scrap-heap.  In workflow situations, QC includes means ensuring that the requirements information is clear, comprehensive and unambiguous.

4. Next, optimise the bottleneck.  If it’s a machine, ensure that it’s powered up first and powered down last and has the least idle time during the day.  If it’s a person or team, defragment their time (blog on this coming shortly), remove unnecessary duties from them and ensure that they don’t waste time waiting for work.

Optimising the bottleneck is about throughput efficiency and effectiveness – every throughput gain made at the bottleneck is throughput gained for the entire system.  The primary options are:

1. Reduce IDLE or WAIT TIME.
2. Reduce SET-UP TIME.
3. Reduce PROCESSING TIME (or Improve processing speed).
4. Reduce REWORK TIME.
5. Prioritise workload

5. Step 5 is, counter-intuitively, to choke the in-flow into the system to match bottleneck capacity.  This may seem hard at first – especially in environments where all work has to be done – and has to be done within an SLA, but there are many ways.  If you can’t find a way, skip to step 7 then come back to 5 and 6.  Ask for help in the comments, if you like.

6. Next, introduce a buffer between QC and the bottleneck, to ensure that the bottleneck isn’t starved by upstream glitches (like machines going down, running out of materials or staff member illness).  The size of the buffer is determined by how long it will take for the upstream glitch to get sorted.

7. Step 7 is to get upstream resources (who have extra capacity) to pre-process items and downstream resources to post-process items to save bottleneck time, even if it costs non-bottleneck time more than it saves the bottleneck time.  (Non-bottleneck time is free and every second saved at the bottleneck is saved/gained for the entire system, so costing a non-bottleneck an hour, to save the bottleneck 15 minutes may seem inefficient, but is actually very worthwhile: we get an extra 15 minutes at the bottleneck, for free.)

8. Finally, we can add extra resourcing/capacity at the bottleneck.  It doesn’t need to be as productive or efficient as the existing bottleneck is: any throughput added at the bottleneck is added directly to the entire system.

Note that, at this point, the third resource can’t do pre-processing, because it has reached it’s full capacity and will soon become the bottleneck.  In most situations, we want the bottleneck to remain in the same place, so we need to take steps to ensure that we have extra capacity at this resource.

Pretty cool, huh?  Eli (Goldratt) was a genius!

Click here to watch the video.