Follow Us On Twitter RSSPrint

About ERP Systems

Ready to convert your legacy data?

Michael Roman - Tuesday, July 08, 2014

Ok.  You educated the company about what an ERP System is and is not.  Everyone now knows the difference between a Master Schedule and a Master Production Schedule.  They have reviewed and addressed the non-value added activities in their processes.  They used those processes to define a set of vendor scripts and the company found the best fit for those new processes from a list of potential ERP vendors.  The contracts are signed the kick-off meeting is over and user training is finished.  Now you can convert your legacy data.  Is that correct?  Well, maybe not.

Have you cleaned up that legacy data?  How many part numbers do you have for a 12” by ½” Standard Thread Bolt?  You hope that there is only one per material type.  Nevertheless, there is also Part 11205 - 12” by 0.5” Standard Thread Bolt.  There is also Part 1205 - 12” x 0.5” Bolt with Standard Threads and also Part 112005 -  twelve inch x ½ inch standard thread steel bolt.  You also looked at your vendor list and you see Jones Plumbing and Supply, Jones Plumbing, Jones Plumbing Supplies, and Jones Supplies.  Strangely, they all have very similar addresses like 1225 Oak, 1225 Oak Street, and 1225 Oak St.  You find some of the same problems in your vendors.  You check the AP Terms and see a Net 10, a Net 10%.  Do you still think it is time to convert your data?  Where else should you look?

You can ignore those problems and choose the one Part, Vendor, or Customer most often used, but what happens if there are balances for some of those abandoned items?  Say Part 1205 has 12000 on hand, Part 11205 has 400 on hand, and Part 112005 has 400000 on hand.  You must not forget you are you are using last cost, and each of those Parts has a different inventory value.  What if there is an AR balance for a customer with three names but the same entity?  Alternatively, what if there are open balances for the same Supplier with three different names?  Do you still think it is time to convert data?  How will you be able to compare inventory values after the conversion to insure data integrity?

That is not an easy question to address and if you ask an accountant they will likely say compare the inventory value between the old and new systems.  So are you going to bring those problems into the new ERP System?  That may be ill advised.  What do you do?  Currently there are not a lot of tools available to remove data duplication for these types of problems.  Often times, companies accomplish control through a set of manual standards. 

We had this problem at a company and quickly addressed the problem for the parts file by creating a description definition template.  The client had 30 Engineers in the company and each was responsible for product development.  Our implementation time line did not allow time to spend attempting to reach consensus in that effort, so the VP of Manufacturing made a command decision.  He defined by material type for our source materials (steel, titanium, aluminum, etc.) and by function, bolt, screw, washer, nuts, etc.

We also created a set of database rules that looked at how we defined supplier and vendor addresses and applied processing rules to not allow ST, St, ST, St., Ave, AVE, etc, etc., ETC, ETC.  We spent a good portion of the conversion effort creating those database rules and new screens.  Continuity moving forward was our goal and besides, we had a huge number of Bills-of-Material that needed changing to remove the old parts and use only one version of those parts moving forward.  This whole effort required much more time on the schedule than the original implementation plan had.  Nevertheless, the company management felt that it was time well spent.

How has your organization addressed this issue?  What have you done to clean-up legacy data during the conversion process?


Michael Roman - Wednesday, March 12, 2014

In Honor of Don Frank, CFPIM, CIRM

In 2004, when Manufacturing Practices, Inc. began, I once again called on Don Frank, CFPIM, CIRM a mentor and dear friend for guidance and joint business opportunities.  We began working on a book and several seminars to help focus the book.  Unfortunately, Don’s health very rapidly failed and I lost him to the ages.  He wrote this piece several years earlier and delivered it at an Atlanta APICS dinner meeting in the 1990s.  He told me to use it when the time was right.  After engaging in a recent discussion on LinkedIn about Part Numbering schemes, this seems the appropriate time.


Let’s go back to basics when we talk about part number attributes.

  • First, part numbers are the data elements or objects that enable us to separate each part from all others as we, in design and operations management, communicate information to each other.
  • Second, the part number enables us to access all the data elements associated with any part in our systems, validating its uniqueness and ensuring we are processing the part we intended. 
  • Third, a part number, assigned to a document, such as an inspection or test report, should appear on bills of material. 
  • Fourth, construct the part number in the simplest lean manner—a pure, sequential numeric form. A good rule for part number length is to add one digit more than the maximum conceivable  number of parts that will ever be in the system. With just eight digits, we can define 99 million unique parts!

People who object to this principle are mostly holdovers from punch card days when, because of the space limitations on the cards, putting intelligence into part numbers.  That perception was the thought that it is necessary for part recognition. Experience, which goes back more than 50 years, was that, even with the limitation of 78 usable columns in a punch card, we could rely better on good part descriptions, rather than remembering the part number, to communicate for what the part number stood.

One of the lessons learned early was to make the part number and drawing number identical, saving a critical amount of space in the part record and making configuration management via revision codes much simpler. We increased the length of the part number to 10 characters, left justified, with the format nnnnnn-nnn, where the first digit represented the drawing size (1 for A size, 2 for B size, etc., so we knew where the drawing was filed). The dash and last three digits we reserved for use with tabulated drawings where several parts represented on the same drawing. An example of this was a set of heat sinks, all made from the same extrusion, but with different lengths, hole patterns, and inserts.

Today's part master databases, with a hundred or more data elements or objects associated with any part master record, enable us to find and visually determine the uniqueness of each part right at the workstation. Original drawings are most often digitally stored rather than on paper. String searches are quick and effective, zeroing in on the part in question in a matter of milliseconds. Just clicking on the part number gives access to all the needed information. We can even hyperlink to a 3-D drawing of the part if necessary.

Highly visible good descriptions will eliminate any excuse for the extra non-value-added task of establishing and maintaining part number coding systems. Descriptions should have two segments—a generic standardized family word description followed by a modifier that differentiates each of the parts in the family. Examples: stainless steel passivated cross-recessed machine screw 10-32 x 1; film fixed resistor 1200 ohm ½ watt 1%.

The first exercise in standardizing part descriptions resulted in reducing the number of parts to support the master schedule from about 5,000 to about 450. The cost savings actually paid for the budding inventory management system.

Another lesson learned was never to use the supplier’s part number as the internal part number because it is too restricting. If you have to change supplier or add an alternate, you create another part number even though the parts are truly interchangeable. Today’s systems allow multiple entries of supplier, supplier part number, and even supplier price against any part number.

Finally, there is still a huge configuration management gap out there because engineering mindsets and product lifecycle management part master data use revision code, and our enterprise resources planning systems use effectivity by date, lot, or serial number.

Here is a word of caution. Do not arbitrarily change existing part numbers when upgrading or implementing new information systems. There is too much engineering, marketing, sales, customer, and supplier documentation out there with embedded legacy part numbers to justify making this type of non-value-added change. Set up a dual-key (alias) system so the system can respond to either old or new part numbers. However, do not allow the sins of the past to perpetuated in newly generated part numbers—use the simple, numeric, and sequential scenario.

Frankly, the only reason we have to put up with long, heavily coded part numbers today is tradition. All new parts generated should have simple, short numeric part numbers. After all, it only takes at most eight digitsto create 100,000,000 Part Numbers! Lean thinking demands we take this approach to intelligent part numbering.

Big Data

- Wednesday, December 12, 2012

In the early part of his career as an aviation engineer with GE, John Bosch pioneered the standardization of gauges on aircraft, solving the problem of too much visual diversity with instrumentation in the cockpit. The way Bosch explains it, all military and commercial engine instrumentation was procured from different manufacturers who built both the transducers and indicators. What the pilot saw in the cockpit were very different displays with needles pointing in all different directions, and these different appearances made it hard at a glance to see the situation of the airplane. Working with the chief pilot of Pan Am, the customer, Bosch basically simplified the complex, creating a system of displays that all spoke the same language. So finally, the customer could see the situation at a glance.

Sound familiar? Isn't that what happens in a manufacturing operation that relies on different data sets?

Wikipedia, now ranked the #5 website in the world, defines the information technology term of Big Data as “a collection of data sets so large and complex that it becomes difficult to process using on-hand database management tools”.  These data sets are important for companies to analyze and understand if they want to make decisions based on the best -- and most accurate -- analysis. Fortunately, a number of toolsets exist to extract and perform data analysis that helps organizations -- small, medium and large -- make quality decisions when dealing with future management issues.

But here's the catch. For the information to be useful, the data sets must be accurate, and after nearly three decades of assisting small and mid-sized manufacturing and distribution organizations, I've learned that basic and vital company data are, very often, NOT CORRECT. Usually it's not a matter of someone making mistakes, it's more a matter of different systems -- like Bosch's system of displays -- that don't speak the same language and the information just doesn't compute, as they say.

Programs that clean-up data are usually a starting point for performing information improvement projects. A recent article in IndustryWeek discusses some of these issues. The article identifies integration issues, accessibility issues, and the inability to analyze meaningful data.  The article also stresses that data analysis needs not be restricted to the Information Technology group but pushed down to the proper user.

Our consultants see the same issues and more often than not, there are bigger issues with data accuracy. Part Number units of measure problems, inaccurate sales order delivery dates, incorrect plant throughput values, Bill-of-Material quantity per values and sometimes, inaccurate “Ship-To” addresses are the source of these errors.  These data must be accurate to be able to produce information that the organization can trust to assist with making proper decision about the future.

John Bosch and the engineers who worked with him simplified the complex in a way that probably has saved lives. I believe that organizations who understand how to improve their Big Data will have a tool that help save their bottom lines. We have written a number of articles that suggest methods to address the basic system data errors.  One article, The Future Ain’t What It Used To Be, recommends data analysis as a useful tool to address company and customer concerns in a proactive manner. Another article discusses how your ERP System can create the Key Performance Indicators to measure employee and department conformance to company goals.  Using KPI in Your Business has had a large readership and references one of first uses of KPIs at Ford Motor Company. Henry Ford was a stickler for measuring, and used measurements to understand whether process were in control or out of control. Like Bosch, he figured out how to simplify an otherwise complex production process.