Balanced vs. Unbalanced processes

This last week, I tested a game proving the advantage of unbalanced processes over balanced processes after reading the Goldratt Institute book Velocity, co-authored by Dee Jacob, Suzan Bergland and Jeff Cox, the co-author of The Goal. I have been trying to find a simple means of illustrating the advantages of combining the different manufacturing theories for years, and was intrigued and eager to test this. I managed to convince five Supply Chain team members to play the game so that I could satisfy myself that it works.

The group started the game with  and were subdued, not really knowing what to expect, or what was expected of them. The curve ball I threw them was that they were all on the same team and not competing against each other: that everyone wins, or everyone loses. We made some mistakes in the first-round while everyone developed an understanding of the rules, fortunately this didn’t affect the overall result, so I didn’t repeat it.  The team slowly thawed as they related experiences in the game to their experiences in different departments in the organization, and finally were cheering each other on to success.

The game has three rounds with each player having 20 turns and we set ourselves a conservative target of 60 batches. Each player follows the clear set of rules, so the only external factors affecting the game are variability and dependency. A lesson learned in the game is that a player can only roll the dice once per turn, no rolling until we get the number we want to see.

assembly-line-balancing-3In round 1, we simulated a balanced line where each step has equal capacity, represented by each player having 1 dice. Variability is exhibited by the range of numbers that can be thrown on the dice and dependency is the affect that a preceding step has on a succeeding step. We managed to deliver 58% of the batches and our WIP decreased by 10%.

 

Observations

  1. Variability has a huge impact on the value chain, average roll of the dice was 3.4, yet only 2.2 pieces moved on average.
  2. Although the line is balanced, the “pieces moved” number drops through each step in the process clearly showing the impact of dependency and you can see the effect of “starvation” on P4 and P5, where their throughput was significantly reduced.
  3. The throughput to WIP ratio is 1.94.

In round 2, we simulated an unbalanced line, and selected 1 player to represent the rate defining step. Each of the other players were given 2 die, representing additional capacity. The result of this round was startlingly different, the team delivered 115% of the promised batches and WIP increased significantly by 435%. The majority of those batches, 55 in total, built up ahead of the constraint. What is startling, is the improvement in the number of batches delivered to market. The downside is the increase in WIP which will lead to congestion in the real world.

Observations

  1. Variability still impacts on the value chain, average roll of dice 6.4, and 4.6 pieces moved on average. The constraint acts as a filter for the downstream steps.
  2. Although the constraint at P3 played close to the average at 3.4, the average for the value chain was 4.6 and the number of pieces moved by P4 and P5 was 3.5 as a result of variability.
  3. The throughput to WIP ratio is 0.79, despite us beating the delivery target.

In round 3, we made one more change, we took the dice away from the “Planner” and the constraint dictated the number of batches issued into the process. Again, the results went against everyone’s expectations, we delivered 107% of planned batches to market and the WIP dropped by 20% as a result of tDBR 1he link between the constraint and the planner.

Observations

  1. Variability only affected the constraint because the other steps in the process all have excess capacity.
  2. Despite the effects of variability and dependency, the throughput was well balanced although lower than the average of 3.5.
  3. The throughput to WIP ratio is 4, meaning that we reduced the lead time per batch significantly, reduced the overall WIP and surpassed our target of 60 batches.

In summary, the game is a very good example of the effects of Constrained planning and feeding production at the rate determined by the constraint, a technique within the DBR 2Theory of constraints called Drum-Buffer-Rope. People continually strive to balance processes in the hope of improving throughput, yet this result shows that the opposite is true, focusing on several areas simultaneously dilutes attention and adds to the workload and stress levels, whereas focusing on the constraint concentrates attention where it adds the greatest benefit, reduces the workload at unconstrained steps and frees up staff to work on batch documentation in real time.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s