setting acceptance criteria for · pdf fileto set acceptance criteria for our validation to...

27
SETTING ACCEPTANCE CRITERIA FOR VALIDATION

Upload: phungliem

Post on 18-Feb-2018

218 views

Category:

Documents


5 download

TRANSCRIPT

SETTING ACCEPTANCE CRITERIA FOR VALIDATION

What Are Acceptance Criteria?

In terms of validation, acceptance criteria establish that your validation will meet it’s intended purpose.

What Validation Acceptance Criteria is Not Not a measure of product quality!!! Not simply a checkbox on the road to product

launch.

The Concept of Variance (and why it is important)

Variance is the natural tendency of a system not to end up in the same place every time (my definition).

In other words, it is what happens when I test the same sample 100 times and get 100 different answers.

The Concept of Variance

Sources of variance Equipment Location People Measurement system Materials

Understanding the variance in your system is KEY to setting appropriate validation acceptance criteria.

The Concept of Variance

In order to understand the variance in your system, you must completely (or to the best o your ability) characterize your process or test method.

The Concept of Variance

Normal distribution Results are spread

evenly around a target. There is no “bias” in the

system

The Concept of Variance

In validation, your variance should be “normal”.

If you find that there is a “bias” in the data generated out of characterization, the system might need a bit of adjustment.

Non-normal variance can be accounted for in validation.

Setting Acceptance Criteria

Now that we have an understanding of variance…

And we have characterized the variance inherent in our system…

We can begin to set acceptance criteria for validation.

Setting Acceptance Criteria

There are many methods of setting acceptance criteria for validation. I am going to focus on one in particular This is using Capability Indices (process

capability) as a measure.

Process Capability

Ok.. Now that we know how to measure variance, how do we use it in validation?

How do we know if the variance in our process or test method is low enough to give us good product or tell us reliable test results?

This is where Process Capability comes in.

Process Capability

Using the concept of capability, we are able to set acceptance criteria for our validation to ensure that our test method can “see” the variation it needs to or our process can produce product in a reliable manor.

Process Capability

Process Potential (CP) Measure that compares

Allowable Process Spread with Actual Process Spread.

USL and LSL are the upper and lower specifications for the product

6 SD (6 sigma) gives you 99.73% degree of compliance + 3.0 SD on either side of the

mean

The larger the Cp, the greater the potential of the process to produce product within spec. limits

𝑪𝑪𝑪𝑪 =𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨 𝑷𝑷𝑷𝑷𝑨𝑨𝑷𝑷𝑨𝑨𝑷𝑷𝑷𝑷 𝑺𝑺𝑪𝑪𝑷𝑷𝑨𝑨𝑨𝑨𝑺𝑺𝑨𝑨𝑷𝑷𝑨𝑨𝑨𝑨𝑨𝑨𝑨𝑨 𝑷𝑷𝑷𝑷𝑨𝑨𝑷𝑷𝑨𝑨𝑷𝑷𝑷𝑷 𝑺𝑺𝑪𝑪𝑷𝑷𝑨𝑨𝑨𝑨𝑺𝑺

𝑪𝑪𝑪𝑪 =𝑼𝑼𝑺𝑺𝑼𝑼 − 𝑼𝑼𝑺𝑺𝑼𝑼

𝟔𝟔 𝑺𝑺𝑺𝑺

Process Capability

Cp gives us a measure of how “good” my window is.

Process Capability A general industry standard is that at a

minimum: Cp = 1 for in process testing or processes Cp > 1.33 for final release

So how do you use this in practical manor to set acceptance criteria for validation??

Using Process Capability

If:

Then:

This is the maximum SD your process or test method can tolerate. Just plug in what Cp you need.

𝑪𝑪𝑪𝑪 =𝑼𝑼𝑺𝑺𝑼𝑼 − 𝑼𝑼𝑺𝑺𝑼𝑼

𝟔𝟔 𝑺𝑺𝑺𝑺

𝑺𝑺𝑺𝑺 =𝑼𝑼𝑺𝑺𝑼𝑼 − 𝑼𝑼𝑺𝑺𝑼𝑼𝟔𝟔(𝑪𝑪𝑪𝑪)

Using Process Capability

When you run your validation, if the variation (SD) is less than or equal to what will give an acceptable Cp, you pass.

If not…

Using Process Capability (a few more points)

One Sided Specifications No spec range (exactly) Use one or the other

depending if you have a lower spec or an upper one.

𝑪𝑪𝑪𝑪(𝑨𝑨) =𝑼𝑼𝑺𝑺𝑼𝑼 − 𝑿𝑿�𝟑𝟑 𝑺𝑺𝑺𝑺

𝑪𝑪𝑪𝑪(𝑨𝑨) =𝑿𝑿� − 𝑼𝑼𝑺𝑺𝑼𝑼𝟑𝟑 𝑺𝑺𝑺𝑺

Using Process Capability (a few more points)

Cp does not account for “Off-Aim” Performance

In this case we use an index called Cpk k is shift factor that

measures how “off” we are.

Using Process Capability (a few more points)

If Cp = Cpk, the process is centered..

Even if the process is not centered, if Cpk acceptable, it is “OK”. You would strive for a

centered process.

𝑪𝑪𝑪𝑪𝑪𝑪 = 𝒎𝒎𝒎𝒎𝒎𝒎�𝑼𝑼𝑺𝑺𝑼𝑼 − 𝑿𝑿�𝟑𝟑 𝑺𝑺𝑺𝑺

,𝑿𝑿� − 𝑼𝑼𝑺𝑺𝑼𝑼𝟑𝟑 𝑺𝑺𝑺𝑺

Using Process Capability (a few more points)

Cp = Cpk if the process is centered. k is a measure of how far off center the

process is: The Cpk index revises the Cp index to account for

the location of the process in the equation.

𝑪𝑪 =|𝑻𝑻𝑨𝑨𝑷𝑷𝑻𝑻𝑨𝑨𝑨𝑨 − 𝒙𝒙�|

�𝑼𝑼𝑺𝑺𝑼𝑼 − 𝑼𝑼𝑺𝑺𝑼𝑼𝟐𝟐 �

𝐶𝐶𝐶𝐶𝐶𝐶 = 𝐶𝐶𝐶𝐶(1 − 𝐶𝐶)

𝐶𝐶 ∗ 100 = % 𝑂𝑂𝑂𝑂𝑂𝑂 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶

Difference Between Cp(k) and Pp(k) Easy Definition: Cpk is the index in the short term Ppk is the index in the long term

Cpk looks at “single” or rational subgroup and the SD is calculated using only that data

Ppk looks at the process over time and the SD is calculated from all accumulated data

Difference Between Cp(k) and Pp(k) Rational Subgroups A rational subgroup is a group of measurements

produced under the same set of conditions. Subgroups are meant to represent a snapshot of your process. Therefore, the measurements that make up a subgroup should be taken from a similar point in time. 𝑃𝑃𝐶𝐶𝐶𝐶 = min �

(𝑋𝑋� − 𝐿𝐿𝐿𝐿𝐿𝐿)3𝑠𝑠(𝑜𝑜𝑜𝑜𝐶𝐶𝐶𝐶𝑜𝑜𝑜𝑜𝑜𝑜)�

, �(𝑈𝑈𝐿𝐿𝐿𝐿 − 𝑋𝑋�)

3𝑠𝑠 (𝑜𝑜𝑜𝑜𝐶𝐶𝐶𝐶𝑜𝑜𝑜𝑜𝑜𝑜)�

Difference Between Cp(k) and Pp(k) If each curve represents

a lot or batch: Cp(k) measures the

capability of the process to produce a conforming lot

Pp(k) measures the ability of the process to consistently produce a conforming lot over time

Therefore: Cp(k) is short

term, Pp(k) is long term

In Summary Using process capability as an acceptance

specification is an easy way to determine if you test method or process is acceptable for use.

If you meet your predetermined process capability acceptance specification, you have assurance your test method or process will meet your needs.

Thank You For Your Attention

My Contact Information: Alan Golden Senior Quality

Professional Abbott Molecular [email protected] 224-361-7159