Operations Science Institute

Avoid “Best Practice” Trial-and-Error Errors

Avoid “Best Practice” Trial-and-Error Errors

Small company performance improvement meeting

 

You want your business to be lean but best practices that worked well at other companies don’t translate to your company. Here is described inexpensive, science-based analysis to successfully improve performance and avoid the trial-and-error errors of “best practices.”

 

I was recently asked to measure production flow efficiency at a small company in the automation sector in Northern Italy. The purpose was identification of improvement actions primarily focused on reduction of cycle time[1] while maintaining, or increasing, throughput. This is a summary showing the quickness and efficiency of operations science analysis. It works for any size company.

 

Operations science can be quickly applied with minimal cost at companies, large and small, in any industry. In a couple of hours on-site, we collected the necessary data and applied the Operations Science Absolute Benchmarking[2] technique, which is a simple, rapid analytical simulation of a process.

 

PROCESS CENTER EFFECTIVE PROCESS TIME (hours/piece) EFFECTIVE PROCESS RATE

(pieces/hour)

NUMBER OF RESOURCES PROCESS BATCH SIZE (pieces) TRANSFER BATCH SIZE (pieces)
Assembly 0.167 5.99 1 1 1
Burn-in 0.171 5.83 1 35 1
Enclosure 0.133 7.52 1 1 1
Testing 0.177 5.65 1 1 1

 

Current Performance

– Throughput = 4 pieces/hour

– WIP = 30 pieces

– Push scheduling control (i.e., without WIP limitation)

 

An initial analysis of the capacity and utilization of each process center (see graph below) indicates that the bottleneck is Testing as it is the process center with the highest utilization rate, 71%.

 

Utilization = rate of arrivals/effective process rate

For Testing, the calculation is 4/5.65 = 71%

 

 

process center utilization graph

 

Note that the utilization levels of Assembly, Burn-in and Test are all similar, they are nearly balanced. This balancing makes optimal control more difficult as variability effects will cause the constraint, also called the bottleneck, to move between the three process centers. This is a great situation in which to use a CONWIP (constant work in process) control protocol. The reason is that CONWIP allows WIP to flow naturally in the process. As variability causes the constraint to change, WIP will flow naturally to the constraint and thereby provide better performance than controlling WIP at each process center (classic kanban) and better than Theory of Constraints drum-buffer-rope WIP control because the constraint does not have to be fixed. See Pound, Bell, and Spearman. Factory Physics for Managers: How Leaders Improve Performance in a Post-Lean Six Sigma World. McGraw-Hill Education. Kindle Edition, for more description of CONWIP.

 

The Operations Science Absolute Benchmarking technique yields the following performance curves for the process:

 

operations science performance curves

 

GRAPH LEGEND

PWC TH – Practical Worst Case Throughput

BC TH – Best Case Throughput

Est Actual TH – Estimated Actual Throughput

PWC CT – Practical Worst Case Cycle Time

BC CT – Best Case Cycle Time

 

The minimum value of the CT is the Raw Process Time (RPT), the horizontal portion of the blue solid line. In this case, we set RPT equal to the sum of the effective times of each phase (RPT=6.48 h). Note that the CT values on the graph are the blue lines and data points and their values are read from the right axis.

 

The throughput values on the graph are the red lines and data points and their values are read from the left axis.

 

Using the Estimated Actual Throughput, we see that we can increase throughput of the line by increasing the WIP in the line. However, through Little’s Law, we know that increasing WIP will also lead to increased cycle time. The question then becomes, what is the best combination of throughput and cycle time that we want. This should be determined by demand.

 

If demand is 10% higher than the current 4 pieces/hour throughput, we will have to move our CONWIP level to 40 pieces (See hollow points on the graph). This provides a throughput increase of 10% to 4.4 pieces per hour and, through Little’s Law again, we see that the cycle time increases to 9.1 hours from 7.5 hours.

 

This simple example demonstrates the ability of operations science to adapt even to small companies, providing major improvements in their ability to successfully control operations. Many useful observations are gained from the analysis.

 

1. The initially stated purpose is to reduce cycle time while increasing throughput. However, the analysis shows that current cycle time is near to best possible. Best possible CT is RPT (6.48 ~6.5 hours). Current actual CT is 7.5 hours. Is it going to make a difference to the business if CT is decreased from 7.5 hours to 6.5 hours? Yes, it is a 13% reduction in CT but do the company’s customers care about a reduction of CT to 6.5 hours when the company ships once per day or once every couple of days? A lean effort to reduce cycle time is not advised here, thus saving significant effort and expense.

 

2. More important are considerations of throughput. Higher throughput, assuming the demand exists, means the company can achieve higher revenue.

 

3. If demand does not justify higher throughput to achieve higher revenue, higher throughput may be used to reduce cost, e.g., less overtime or fewer shifts.

 

4. If additional WIP is used as the lever to increase throughput, the benefits of added throughput can be weighed against the increased CT. There will also be some increase in working capital with additional WIP, depending on materials cost.

 

5. The operations science analysis can also be used to gauge the effects of changes to the process, e.g., adding additional pieces of equipment.

 

What I have described here is a powerful, practical science-based analysis that can be used by small companies in any business to great effect with minor cost. Consideration of other companies’ best practices can be helpful. However, don’t rely on a trial-and-error approach using “best practices” that may or may not produce desired results. A much better approach is to start with the operations science describing your company’s performance and then adjust where the science indicates improvements make sense for your business.

 

Create an account at login for more free, highly useful resources such as an operations science glossary and case studies available for download. Once your account is created, go to Resources/Articles/

 

Contact the Operations Science Institute at info@opscience.org for more information on how your company, large or small, can benefit from learning and applying the concepts of operations science.

 

Sandro Rizzoli, PhD is a freelance business consultant from Bologna, Italy working with production and service companies in northern Italy. Dr. Rizzoli’s PhD is in Economic Statistical Sciences from the University of Bologna. He focuses on quality management systems and application of operations science/factory physics while helping companies improve their performance. Sandro can be contacted at info@sandrorizzoli.it

 

 

 

[1] Cycle time, as used here, is the time to complete the routing through all four process centers—from Assembly thru Testing.

[2] Absolute Benchmarking is a technique for determining how well a process is performing versus its performance boundaries (absolute best and worst performance limits). See Pound, Spearman, and Bell, Factory Physics for Managers (New York: McGraw Hill, 2014), pp. 81, 84 – 92 for detailed discussion.

Leave a Reply (Login required)

Leave a Reply

Continue Reading