First Article Inspection – A Comprehensive Guide on How to Perform an FAI

First Article Inspections (FAI) are used to ensure that parts off of new or modified tooling or processes conform to the part design requirements.  This even includes changing the location of manufacture! Yes – this can make a difference sometimes (usually due to different equipment, the same equipment setup slightly differently, environmental conditions, different manufacturing standards, etc).  This includes any time the Form, Fit, or Function could be impacted.

When to Perform a First Article Inspection

This inspection is performed on the first pieces (articles) that come off the new or updated process or equipment. Typically the supplier of the part will submit 3-5 pieces with a full dimensional inspection and material certification.  If the parts are rare this may be reduced to a single piece.  Multiple pieces are done to help ensure consistency in the manufacturing process.  The data is then compared with the buyers (the end customer) own inspection to ensure that they have agreement between measurement methods.  If there are discrepancies the two parties work them out by either changing the tooling or process to conform to the design intent or changing the measurement method or changing the design to allow for the deviation.  This is where the trouble can start.

The supplier wants to sell the parts that he has just created tooling/processes for and the customer wants parts that will always work as intended.  This point in time can be where the supplier realizes that the specifications that were originally agreed to are way too stringent (dimensions too tight).  The quoting of the parts on the supplier’s side usually assumes that this project will progress as other similar projects have.  There is a great pressure on the company quoting to deliver a quote in a timely manner. They know from experience that slow quotes usually do not win the work.  So this puts pressure on the suppliers engineer to rapidly quote.  They typically have a CAD model and a print.  

Sometimes they have to quote without CAD or without a finished drawing. These are ROM quotes (Rough Order Magnitude). And these should be subject to change based on the final requirements on the finished drawing.  This does cause tension as the rough quotes can be wildly off due to last minute odd features being added to the design or tight tolerancing requirements.

The finished prints can be very detailed and difficult to think through all the tooling requirements and their consequences.  Sometimes, you can hold a specific dimension if the feature were all by itself but you cannot hold it while at the same time holding several other tightly tolerance features.  Those conflicts usually are not actually seen until the tooling is being designed.  These things are sometimes worked out during customer/supplier meetings to approve the tooling/process design.  But this is also where added scope items creep in which will later cause problems during the FAI.  The pressure for the supplier is to give in to the customers’ requests and if they are not careful they may often do this for free.  The temptation is to think: ‘Who knows – Maybe this time they will be able to do what we have never done in the past?’  The suppliers technical people will tend to fight the changes and the suppliers sales force will tend to want to accept the changes to save the sale – win the job.  Those tooling design review meetings that happen after project kick-off are critical.

Once the first pieces are made from the new process, then the next step in the FAI process begins.  The first articles are supposed to represent what the tooling/process is capable of producing.  However at this point, the supplier has not been making these parts for any period of time and is still trying to figure out times and temperatures and other process parameters.  They find a ‘process’ that yields what they think is a good part and then start a dimensional inspection to verify.  For the sake of argument, we will assume that this is a relatively complicated part and that some of the dimensions do not meet the print tolerances.  The supplier then tweaks his process and re-measures until they have the vast majority of the data in specification.  NOTE that this process may not be a very stable process at this point.  The final process that makes the FAI parts has really only been run once!

Beyond dimensions that are slightly out of specification, there is another category of dimensions that is usually overlooked by the customer.  This is the group of dimensions that are technically in specification but very close to the edge of the tolerance zone.  This can be problematic as you consider the matter of measurement uncertainty.  When you take a measurement with a CMM, calipers or whatever instrument there is a band of uncertainty around the measurement (also called the measurand).  Basically the instrument you are measuring with has a tolerance to it as well.  It will only read as good as it is capable of reading.  Consider a tape measure versus a CMM in a temperature controlled room – which dimensions would you trust more?  FAIs should have a listing of each instrument that was used to make each measurement.  They should also have a listing of what the measurement uncertainties for each instrument is.  3D Engineering Solutions is certified under ISO17025 which requires the estimation and reporting of measurement uncertainties.  I have seen many FAIs not even list the instrument used (or only vaguely list them as ‘CMM #2’).  This is a poor practice and can be somewhat deceitful.  What exactly was the equipment used for the FAI?  Was it even in calibration? Was the calibration done by a certified source?  Further – if you have a dimension close to the edge of the tolerance zone, is it actually in or is it so close that the inaccuracy of the equipment puts this in doubt?

Instruments Used in an FAI

The repeatability of the instruments used to measure the part comes into play as well.  Because of an inspection method’s measurement uncertainty, I have seen suppliers measure the same dimension multiple times until they get a ‘good’ answer.  And because of measurement uncertainty, these final ‘good’ measurements are technically one of the acceptable answers – if not a reliable or repeatable answer.  The person preforming the measurements does have the authority to ignore spurious data.  However, they need to be careful in their judgement. I usually see this in results that just happen to be exactly on the border of the tolerance zone on multiple samples or in readings that are exactly the same (to the micron) on multiple samples.  As a service provider, we are often asked to review others data and independently measure those parts as well.

It is not only the estimated measurement uncertainty of the instrument being used but also the uncertainty of the dimension being measured.  Take radii for example.  These are dimensions that typically have the highest variability. Because you are trying to measure the radius of a circular feature with by less than 90 degrees of that circle (<25%), there is high variability in answers from any instrument.  If you had 60% of a full circle, then at least you would have at least some opposing points that would help stabilize the measurement (60% is the general rule of thumb for high confidence in the measurement of a circular feature).  Since radii have such a small percentage of a full circle, this will add uncertainty to the measurements. I typically see suppliers use radius gages (basically a subjective go/no go gage) to measure these features.  And some times that is the right thing to do because of the aforementioned difficulties.  Another way to handle this particular feature would be to change this from a +/- radius dimension to a profile requirement.  That way the radius is not involved and it just requires the surfaces to be within a specified range of nominal.  Typically, radii are used for clearance or for tooling consideration or fracture prevention in the finished part (creating areas of low stress concentration factors where there would otherwise be a sharp corner).

Sometimes an FAI package requires a Gage Repeatability and Reliability study (GR&R).  This is to be done on the supplier’s measurement equipment and process only!  We have been asked to provide GR&R studies on measurements that we make as a third party supplier for FAIs.  However, this technically not correct and not in the end customers greatest benefit.  The point of a GR&R is to ensure the end customer that the measurement equipment and process used to measure their parts is reliable and repeatable.  If we as a third party are asked only to perform a single FAI and all future measurements and in process checks are done by the supplier, how will the customer know that they can trust the supplier’s measurements?  The idea is to test the equipment that will be used long term for its fitness for measurement.  Additionally since each measurement is unique and is subject to its own issues (consider the radii example from above), a GR&R should be performed for each critical dimensions as defined by the design engineer.  What we might typically see is that only a few relatively easy dimensions are used for GR&R study.  Those dimensions are usually picked on agreement between the manufacturing engineer on the customer’s side and the supplier.  These should be specified by the design engineer.  This is because; he is ultimately responsible for the fit form and function of his design.  I have seen GR&Rs performed, poor results received and new dimensions chosen and substituted instead.  What should happen is that a different measurement method and process should be used.  The purpose is to verify the fitness of the measurement method and not just to receive a good result.

What is the Purpose of First Article Testing and Approval?

The general process of a First Article Inspection is a back and forth between the supplier and the customer.  The supplier should include FAI costs in the overall project costs.  We generally see that there is not enough cost for this in most of the more complicated projects. Because it is the first time that everyone is looking at the parts in detail, every dimension should be measured and reported for each of the parts.

What is Measured in a First Article Inspection

All toleranced dimensions should be measured.  Whether it is tolerance by a +/- tolerance shown directly on the dimension or if it is tolerance indicated by the tolerance block on the print.  If the design engineer did not require it to be specifically dimensioned on the print, then it is not checked.  Many design engineers get around this by adding a general profile tolerance note.  This might read ‘all surfaces are to be profile 0.060” unless otherwise specified’.  This is a catch all to help ensure that that the supplier doesn’t go crazy with a particular area to accommodate a tooling/process feature.  

What is Not Measured in a First Article Inspection

What should not be measured are basic dimensions.  These are the GD&T (Geometric Dimensioning and Tolerancing) dimensions that are enclosed with a box on drawings.  By definition, a basic dimension is always equal to itself!  So reporting a basic dimension is redundant.  What we find in our inspection service is that customers will request that basic dimensions be measured.  Sometimes they are unaware that these should not be measured as they are not fully aware of how to interpret GD&T features.  Sometimes, they want to understand how a particular GD&T features is varying from its nominal (in what direction). Measuring basic dimensions is problematic for a few reasons: 

1. By definition, a basic dimension has no tolerance.  So what do you compare it to?  What defines good and bad?  And if this was not specified early in the process for the supplier – it is not fair and should be considered scope creep. We have had to add hundreds of dimensions in some cases due to this for an FAI.  Each dimension has a cost in time and dollars.
2. Since basic dimensions help establish the location of GD&T features that are controlled by tolerances, measuring them with a made up tolerance is the same as double dimensioning on a print – a very definite no-no.  What happens when one basic dimensions conflicts with another?  What happens when the basic dimension is ‘out’ and the GD&T controlled feature is in spec?
3. It is not clear if the dimension is to be the shortest distance between features or if it is the distance between features following a particular datum.  Let’s say that you have 3 holes that are supposed to be in line with each other (by apparent print intent) and one of them is not in line with the other two.  Let’s further say that there are 2 basic dimensions shown on the print.  One between the first and second hole and one between the second and third hole.  Do you measure the shortest distance between each set of holes or do you measure a linear distance based on some edge of the part (that may or may not be a datum)?  For this example, it is likely that there is already a true position callout for the holes. In that case it uses a series of datums to align the effective measurement. Note only one specific datum!  So, whichever option you would have picked would technically be incorrect.  You could chose to measure this using the same datum reference frame as the holes.  However, sometimes this is not clear as there may be multiple datum reference frames used for the same features.

The print dominates in discussions on FAIs.  It is the one source that is supposed to communicate the design intent of the part.  If there are 30 instances of a dimension, but the print only calls it out 1 time, then only 1 of those is checked and it is the specific one called out.  This makes the print checking process very important.  I find that this checking step is most often left out or brushed over quickly.  A well written print communicates the design intent and leaves little to argue about later.  The relatively small amount of time you put into communicating your design intent can save much time and money later.

How Many Parts Should be Used for an FAI?

The customer must decide how many parts are to be used for the FAI and if they should be done on multiple setups or if one is acceptable. I typically see between 3-5 parts being used for an FAI.  The more parts you use the better confidence you have in the results.  However, the more parts you use means more cost and time.  Customers tend to get very antsy when physical parts are available and forget that it takes time to program and measure every dimension.  It is very rarely that I have seen an FAI that is done on more than one process setup.  This is due to the fact that the tooling/process is new and no one wants to set up the process twice and show results from multiple setups.  There is extra error due to repeatability of most processes.  The measurement results look better to have all the parts come from one run or one lot.  You have to decide if the extra setups are needed for your particular part.

All of the data for each of the dimensions measured should be shown together.  This allows the customer to see each of the dimensions from each of the sample parts beside each other vs. having to look through multiple reports.  Many customers like this data in spreadsheet format.  They may ask you to add conditional formatting to highlight when specific dimensions fall out of the tolerance specification.  If they don’t ask, go ahead and do it anyway.  It adds to the positive flow of communication.  NOTE that you would typically be doing this without regard to measurement uncertainty (it is a best practice to add a note to your FAI indicating this).  You would only be showing how the reading you measured compares with the tolerance zone. This goes for not only variable data but attribute data as well.  If there is a print requirement for part number marking, the result might be ‘yes’, ‘present’ or ‘true’.

Each dimension or print requirement in your report should be ballooned on a copy of the drawing and referenced in your report.  If there are multiple instances of a dimension, then they should be represented by separate numbers or by adding a letter.  For example if there are 4 holes in a pattern, they might be specified as 1, 2, 3 and 4.  Or they might be represented as 1a, 1b, 1c and 1d.  Missing the number of instances of a dimension is a common mistake in quoting inspection work.  Be careful when counting.  Sometimes the multiples are shown in the title for a section view.  For example, it might read ‘Section A-A Applies in 6 places’.

Key Parts of Your FAI

Your FAI report should include a ‘methods’ section at the beginning.  This is a spot to allow you to describe any special considerations, notes or measurement issues that might apply.  All too often, I see FAIs with simply a list of measurement results only.  Most standard reports have a place for the inspector to list their name and equipment used, etc.  However, this is often not filled or only partially filled out.  If there was a special setup that was required to measure a particularly difficult feature, it should be described and pictures added for clarity.  Adding this establishes more credibility in the measurement results. It could be that the customer had always measured this in a different way.  If you explain non-standard measurement techniques, it goes a long way to build trust and confidence in your results.

On the customer’s side, remember that you chose this supplier because of multiple factors.  For complicated parts, it is often impossible to meet every dimension for a reasonable price (or the price you agreed to).  The supplier has a desire to do the best job possible with the technologies and resources they have.  However, they must be able to do this efficiently and economically.  I have seen end users of components ‘nit-pick’ over the slightest discrepancy and send the supplier into a seemingly never ending loop of tooling modifications only to later realize that the design engineer would have been fine with modifying the print instead!  Note that every time you make unnecessary changes to a tool (welding and cutting for example), you decrease the life of the tool which would hurt you in the long run in time (remaking and verifying a tool) and money.  This is what makes ‘tolerance stacking’ so valuable to the design engineer.  They are often given little time to do proper tolerance stacking.  However, if they have done stacks for their product and find a feature that cannot be held to the requested tolerance, then they will be able to see what impact it has on the design and could even move tolerance from another part in the stack to this feature to help accommodate.

It is when the FAI process starts that much of the tension for the project happens.  The suppliers are anxious to sell the project and start making money on their investment and their customers are anxious to have functional parts to add to their complex and time critical project. 3D Engineering Solutions serves both suppliers and their customers as an independent third party measuring source that is certified to ISO17025. Our independence inspires trust in the reliability of our data.  We can stand at an arm’s length in the FAI process between the suppliers and their end customers and help resolve conflicts.

Please contact us for all of your FAI needs.