From what I’ve seen, most people let their filing cabinets and computer desktop folders grow organically without much of a strategy. And they end up suffering for it needlessly.
Getting organized is one the simplest things you can do to reduce the stress of work and improve your overall productivity.
The goal of a data management system is to ensure that you can find the information you’re looking for, even if you’re looking for it years after its creation. Strategic organization of physical and digital data gives you quick access to what you need when you need it, and makes the most of your time and resources. If your system doesn’t do that then it’s time for an upgrade.
The core reason a good data management system is important can be summed up with this universal truism:
When I designed the Bicycle BoosterPack toy I initially did a terrible job managing all the information associated with it. I focused solely on the technical challenges and ignored good data management procedures for the sake of moving quickly.
Very soon I had a huge folder of images, video segments, 3d model versions & variants, the bill of materials, assembly/testing/experimentation notes, fabrication notes, drawings, wiring diagrams, and all the other data I would use to write up a long and technical set of build instructions.
At the end I got some model versions confused and I realized that I had a problem that needed to be dealt with immediately and permanently before moving on.
Establishing a real digital organization system after you already have a mess of information is tough, but I think it can be done in three basic steps.
Discard excess documents, then decide where to keep things, then back it up.
STEP 1. CLEAN UP
One thing hoarders do not understand is that storing anything, be it physical objects or even information, has a real and measurable cost. In this context the cost of storing useless information is the incremental obfuscation of stored useful information.
Go through all your documents and get rid of unnecessary crap. Do this by category rather than by location so you can more easily mentally inventory everything you have and quickly decide what is garbage. When possible consolidate similar documents.
During this step it’s easy to get drawn in to working on a side task but don’t get distracted. Clean up must be completed in full before moving on. If you can’t decide immediately if something should go I have found that keeping an “Archives” folder to be a handy temporary storage location.
STEP 2. Establish rules for storing things.
Thoughtfully organize the remaining information via ‘The Five Hat Racks of Sorting’. That’s a metaphor to represent the five different ways information can be organized, which are:
- Location: Sort by spatial reference
- Alphabetical: Sort by alphabetical sequence
- Time: Sort by chorological order
- Category: Sort by specific similarity
- Hierarchy: Sort by magnitude
For example, I have WAY too many digital photos of my wife, baby, and dog. And I plan on taking many more over time so the most useful way to sort my pictures is by when they were taken.
I have a series of folders with numerical names that represent the year and month they were taken, this month’s photos will go into a folder called “16_12”. That way when I sort the folders by name they automatically go into chronologic order.
Another example: The documents I create for EngineerDog products are sorted using categories. Each product has its own folder with 3 subfolders called: ‘Media’ (for photos, art, videos), ‘Model’ (All 3D & 2D CAD), and ‘Supporting Documents’ (for printable packing list, product label, article notes, & BOM).
Whatever you do use plain language and logical descriptive names for your folders. Avoid ambiguity, acronyms that aren’t in the dictionary, and weird permutations of words (such as sk8 instead of skate, scrw instead of screw).
Nesting folders is fine but try not to create a labyrinth of mostly empty folders. Mathematically speaking, for any given data set there are always a greater number of ways to organize the datums than there are number of datums in the set. For example, in the US less than 4,000 companies are actively traded in NYSE or Nasdaq, but there are nearly 10,000 mutual funds consisting of different combinations of these stocks.
The Really Tricky Part
Sorting can be tricky because the same information can often logically be sorted in different ways. It’s up to your puny little brain to figure out the best way to map file structures depending on your specific needs.
In cases where you find multiple logical places to put things it’s OK to leave a shortcut to the master file in the ‘other’ place. But never have two copies of the same file!
Keep in mind how large your system is likely to get over time and how many people will be using it. Beyond a certain threshold the 5 hat rack method breaks down because too many things can have too many logical homes. In this case you have to start relying on a search program and file tagging system instead.
STEP 3. Backup
Preservation of the information is a critical final step for any digital data management system. Imagine trying to reorganize if suddenly half of everything was missing.
I don’t own any digital data older than 2007. Why? Well I wasn’t very careful with my toys in high school.
I remember videos and photos of my family, friends, dog, and me that are lost forever. Lesson learned, preventing data loss catastrophe is cheap and easy enough that it would be foolish not to take precautions.
$55 gets you a 1TB Portable External Hard Drive, which is an absurdly large amount of data storage for one person. You won’t be surprised to hear that I keep mine in a fireproof safe…!
Special Cases of Managing Engineering Data:
The above discussion is universally applicable organization strategies. Here are some specifics cases only relevant to engineers.
Some companies have sophisticated 3d model sharing systems that have built in version control that allows multiple users to access data at the same time. For a guy in his basement using cheap software having some sort of version control system is more difficult but still a must.
A common method is to attach an X.Y number format to your models to track them. For example if a part is versioned ‘4.5’ the 4 digit is used to indicate a major revision (changes of form or new features) and the 5 digit is used to indicate a minor revision (change in fit/location/orientation of existing feature).
If you copy a part and develop it along a different path for different purposes from the original (changes of function) then it is a variant rather than a version and deserves a new part name starting at 1.1.
- Prototyping Recordkeeping.
When a project calls for a lot of prototyping, with minor variations of similar parts being built and tested in tandem, its super easy to get things mixed up.
When I build things, I save a digital copy of the part or assembly with the name ‘Partname-As built V1.2’ into the ‘archives’ folder so that I can always refer back to it later. It’s good practice to label the physical part with the version number and/or date on it too.
- Bill of Materials.
A BOM is a basic and essential recordkeeping tool for tracking the parts used in an assembly. It takes the form of a list for recording information about every part used in an assembly.
Typical info I like to have in my BOMs are checklists for parts ordered/received, a record of what each component is exactly, how many were used, what is the cost per, where you got it from, and so on. I’ve provided two simple examples for hobby use in the EngineerDog resources page here.
The trick here is knowing that a BOM is a working document intended to accompany an engineering drawing to fully define an assembly. Without a BOM your assembly is missing critical information. Even for a small project or one-time thing you don’t expect to build again, a BOM can still be useful for keeping track of things, even if temporarily.
Hopefully this is one sickeningly obvious to you. This is basic stiff and yet I mention it here because the number of times I’ve been inconvenienced in the past by someone else’s decision not to keep a BOM is even more sickening. It’ a newbie mistake that results in the need to recreate information that had previously determined.
- Part Naming & Numbering Conventions
This one created havoc for one of my previous employers. There was no standard naming convention for parts so when I wanted to find existing options already in the database there were too many possible permutations to search for to find anything! (Their database software was garbage so this was not a trivial problem.)
What one person called a ‘metal screw quarter inch coarse’, someone else called a ‘steel bolt .25” x 20’, and still others called ‘Black Oxide Alloy Steel SHCS ¼-20’’. We would end up creating something new even though there was often a perfect substitute in the system that we couldn’t find. This created a mess for inventory control.
What to do when naming a part: A good generic method is to use the form
Common Pronoun – More specific detail – Even more specific detail
Call it what it is first, describe its function second, and where it is used within a specialized sub-assembly last (if applicable). This is almost completely reverse in order from how you would verbally describe something (in English) but it is best practice from the perspective of searchable names and communicating what things actually are.
Avoid special characters (“,&%$#”), frivolous information, non-numeric attempts to indicate the version, or the name of the project that the part is used on. Some software requires that part names be one word and in this case hyphens or underscores are clean ways to separate descriptions.
- Fan- Axial 55mm
- Bushing- Spring 10mm
- Sprocket- 50T #40
- Cover – Geartrain
- 55mm axial fan
- Boosterpack Wall #2 (This is too generic and will cause mix-ups later.)
- Spring Tension Bushing Newest & Best (what will you name the next version?)
- Gear cover (if I wanted to find a gear this would pop up.)
- 50 tooth Sprocket #40 roller chain (Not clear if part is sprocket or chain)
- Michael’s Gigantic Motor Mount (just no.)
So what is the best naming convention for fasteners? ?Type_Diameter_Pitch_Length_Material_Standard_Finish_Head_Threadclass?
It’s a trick question. There is no universal standard because there are too many slight yet significant nuances among fasteners to make a set of rules to reliably fully specify them with a tiny description. Besides, how many different fasteners are you ever to likely use? You don’t need to define every possible fastener.
If you are doing a one-off hobby project then just use vendor part numbers as fastener names.
If you are a forward looking small organization then realize that is an instance when the 5 hat rack sorting method fails against a good database & search tool. The best option is to keep a fastener specific database based on a ‘dumb numbering’ system. (The opposite of a ‘smart part numbering system’ that attempts to make each digit have a meaning.)
In a dumb number system each fastener should have a numerical name drawn from a sequence of non-significant non-repeating numbers devoted to fasteners. The full specifics of that fastener are then store in a database correlating to a single dumb number. This numerical name unambiguously identifies all the specifics of the part.
Well I’m sweating after finishing up this one so I’ll quit for today. I think the above list provides a solid introduction to the way you should manage engineering data. I hope you’ve found this to be helpful!
This week I posted a new gizmo on the EngineerDog Store! A simple tool born out of necessity, a low filament detection alarm for 3d printers.
After getting this thing developed, tested, produced, and put on the shelf for sale, I’m patting myself on the back for having done the data management thing right this time around.