Data Analysis Term Paper

In our search for takt time, Allison Kennedy, a Memorial University Naval Architecture student, wrote a final work term paper on, "Lean Principles in a Marine and Shipbuilding Engineering Environment."

This detailed analysis used data collected from past projects at Genoa and tested theories of production rates existing in the marine engineering design field. This is something we've been working on as a R&D project and I've referred to it in past blog postings.

While the entire paper most accurately describes the process and findings, the following is a summary of Allison's conclusions and next steps.

It doesn't read like a John Grisham novel but contains some valuable information.

Ken Hogan
Lean Guy at Genoa

"Lean Principles in a Marine and Shipbuilding Engineering Environment"

Conclusions
Implementing lean into a virtual marine and Shipbuilding Company is not a straightforward process. There are no physical objects to count or measure in order to determine the production rate. It is still unknown if a valid production rate can be determined by analyzing accumulated data from past projects. This may be an aspect contributing to the fact that there are hardly any similar companies worldwide that have attempted to implement lean. Since there are no guidelines to follow, the process is deemed as research and development and this adds to the level of difficulty of the problem.
There were a total of six tests completed using the accumulated data from past projects in an attempt to determine a valid production rate. Five of these tests were based on the same idea, which is to use the collected data to find parts per volume and parts per hour value. These general values were used to determine the number of completion hours required to complete future projects. The first step taken was to determine the calculated number of parts by multiplying the calculated parts per hour value by the known volume. Next, the number of completion hours could be found by dividing the calculated number of parts by the calculated parts per hour value.
Tests one through five grew increasingly more detailed in attempt to determine a smaller completion hour variance. The first test was general and determined ppv and pph values from analyzing the entire data set. The second test divided the data based on ship type assuming that similar ship types would have similar ppv and pph values. Test five was the most detailed test in which the data was split into ship type, ship section and ship length. The outliers were also removed in this test. The results prove that as the testing process moves forward, the completion time variance between actual and calculated values becomes smaller. Test five provides the best results having a variance of approximately 23 percent. This variance is still high and therefore the production rate calculated would not provide beneficial results. However, one must note that the results are continually improving as testing continues. This is a positive factor, which gives hope that an accurate production rate can be achieved.
Test six was based on the determination of a volume per hour value from the analysis of vessels of similar types. The idea for this test was sprung from the thought that additional inaccuracies are introduced from the calculation of a number of parts and that this uncertainty was introduced into the calculated number of completion hours. The method used in test six is more direct than that of tests one through five because the completion hours are calculated in a single step. The results of test six are poorer than that of test five. The barge results of test six are compared to the results of test five since test five was completed only on barges. It can be seen that test six produces a variance that is higher by approximately ten percent. This was the first test completed using the volume per hour method, further testing will undoubtedly determine a closer approximation.

Recommendations
There are a number of recommendations which should be taken in order to develop the most efficient and effective method of completion time determination. These are all outlined below for review and consideration:
1. The validity of data should be studies to determine usefulness. As discussed, old data is less valid than newer data for various reasons. A study could be completed to test variations in new and old data and remove any old data which falls out of a certain range.
2. Data capture process should be more accurate. A process should be put in place so that employee time recording must be done at the end of each day. This will decrease the chance that time is allotted to the wrong task and increase the validity of the collected data.
3. There should be consideration put into performing an analysis based on the input of design drawings received from the client. Some details contain less detail causing more PM, set up, modeling etc.
4. An analysis should be done in which the actual hours worked are compared to bid hours. This will give an idea of the variations in bid and actual hours.
5. Additional columns should be added to the ‘Time Estimation’ spreadsheet to ensure the user compares the vessels, which are indeed the most similar. For example the overall length, draft and hull form coefficients should be displayed for each of the completing vessels so that they can be compared with the proposed vessel.Research and development should continue in this area so that an efficient and effective method of determining completion time can be found.

New Year's Resolutions

It's a new year, and the air is full of resolutions.

I just received Jim Womack's latest e-newsletter concerning cadence, and the fact that he has resolved to write one e-newsletter per month.

Personally I have set goals that may very well exceed my capabilities in 2008. Writing this blog once per month may be an example...or may not.

Already, my to-list list is far too long to see an end.

How to prioritize? How to start the lengthy task that we know will take 2 days to complete, when there are other tasks of similar priority that only take one hour and can be easily stricken off the list? How on earth does one organize such a mess? Throw on top of this a Lean Project that will occupy the Lean focus within our company for the next 5 years? With so many items that need attention, focus, teamwork and research, how to pick which Lean project or task to accomplish first?

Cadence mixed with prioritization mixed with low-hanging fruit dangling around our heads...

I believe that this is why we have a tidy little tool we call the Lean Diagnostic. This handy-dandy tool enables us to self-evaluate in an organized fashion, within a variety of topics that span the complete spectrum of running a business.

So as we analyze, it becomes obvious through the low scores, where the weaknesses are, and where the priorities lie.

So as we analyze, we create pull.

Isn't that what a good to-do list should do? It should pull our efforts from us. But the to-do list must be developed intelligently enough that it pulls from us as the customer demands, not in an arbitrary fashion that depends on our moods, or our likes or dislikes. One can compare this to brain "setup" and batching in our brain according to our likes of the day. Every good Lean practishioner knows that batching ain;t good.

Back to cadence. Introduce cadence into the customer pull, and you come up with a way of setting daily goals to strike to-do items off your list. But how do you know how many to accomplish per day if you have not levelled the individual tasks? How about conducting a Value-stream mapping exercise on a to-do list to see which items are wasteful? How about levelling the tasks into equal efforts so that we can set a cadence for ourselves? Not quite Takt, but the cadence that Jim refers to.

So...with these things in mind, I should be able to insert this blog-writing task, which is of equal effort to other items, into my to-do list, set the day, accomplish the task, and strike it off my list once per month. Thus, I too have cadence, and the blog gets done regularly.

This is not the final frontier, but when we can start using Lean lessons and tools to change the workings of our brain, we start making some tremendous hits. To 5S the brain is like finally letting the 5S team enter your office to mess with your drawers. After all...we all know that 5S is good for the other people, but will never work for us.

I think this year I have made some resolutions:
1. Let the 5S team into my drawers.
2. Set a cadence for my to-do list.
3. Get the blog out once per month.

A couple of fun resolutions:
4. Change my voice mail daily, so people know they are leaving messages on a dynamic system, not a useless tool.
5. Run a road race.

Leonard