"Duration" variable in DOE

[p.wozniacki]

Please help me in correctly interpret the output of DOE analysis where one of the variables is Duration. The DOE log reads:

Duration

Range = based on multiple values, simplified to [-1, +1]

How is the simplification method working, and what exactly is Duration of -1 when the packing profile as per Process Settings has been:

0 sec -- 80% Filling Pressure

7 sec -- 80 % Filling Pressure

3 sec -- 0% Filling Pressure

Thanks,

Piotr

[p.wozniacki]

Of course the packing profile provided in my original message is meant as an example only on which I'm seeking explanation of the real "-1" or "+1" settings to the "Duration" variable in DOE analysis (this applies to the Beta 3 of AMI 2012).

I'll appreciate any comments, as they will help me make sure I understand the concept correctly.

Thanks again in advance

Piotr

[WarpGuru]

Hi,

To answer you question, (-1) corresponds to the minimums in the profile table of the DOE builder and (+1) corresponds to the maximums. Since DOE needs a range and that anyway every variable range is mapped to [-1,+1]  inside the solver, it made sense to choose those values.

To isolate a particular state when you have a result between [-1, +1], just apply:

profile point value = average + delta * value [-1,+1] 

or = (maximum+minimum)/2 + (maximum - minimum) / 2 * value [-1,+1]. 

Cheers,

[p.wozniacki]

Thanks a lot, I feel reassured now.

Piotr

[p.wozniacki]

What I’m still not quite sure about is that - will multiple, not just a single, profile steps as DOE variables - I’m getting just one set of (simplified) pressure and duration outputs from DOE analysis. How do I tell which step I should apply them to?

[WarpGuru]

"What I’m still not quite sure about is that - will multiple, not just a single, profile steps as DOE variables - I’m getting just one set of (simplified) pressure and duration outputs from DOE analysis. How do I tell which step I should apply them to?"

Well, I am not sure I fully understand the question but ...

Just to clarify, a profile variable btw [-1,+1] represents a set of profile point values. You can create a variable by applying a min/max to one of the profile points or multiple. Actually, you can even go for a convex shape for the minimums to a concave shape for the maximums. There is really no limitations. The [-1,+1] range will represent the transformation from the minimum shape to the maximum shape.

Here is what I think is the answer to your question, the user interface was designed to allow the creation of two variables from one profile, one for each axis so that variation of pressure could be compared to the variation of duration in one DOE analysis.

Let me know if it answered you question,

Cheers,

[p.wozniacki]

Hi

Thanks for your patience with me, but frankly I must admit I'm still feeling uncomfortable about it 😞

It's kinda embarrassing for somebody who's been working with Insight for so many years, but could you please explain it to me using an example? Like for instance the simple packing profile I quoted in my original post.

Please don't get me wrong - I do grasp the idea (in fact, I used the previous DOE analysis quite often). So, the most "influential" variables are established, and the responses calculated. The optima foe each qualities are found, and their corresponding variables' values listed in the log file... And here is where I feel uneasy: let's say some quality is at its optimum :

with -1 as %Filling pressure

with 1.001 s as Injection time

with 1 as Duration

- why can the Injection time be expressed in seconds, but Duration - as a number from the [-1 +1] range? How should I modify my packing profile if all Duration-%Filling Pressures pairs have been checked as input variables?

I hope what I'm writing makes sense - but am not sure it does at all, because when one has problems understanding something, he usually also has problems explaining his doubts in a clear way 🙂

Something must be wrong with me - perhaps I'm tired?

Piotr

[WarpGuru]

Hi,

No problem, always happy to help,

Let say you have the Duration profile below.

DoE works with numerical ranges i.e. Average [Min, Max]. In the profile below, we could use the last value 10 [8,12] but what if for another profile, the last value was 0 [0,0]. Then, it would not work. The solution was to map the entire group of profile value {0, 5, 10} to one numerical range. The basic form of a design (the method used in DoE) uses range of [-1,+1] and every variable range is eventually mapped to it inside the solver. So for profile, it felt natural to go directly to [-1,+1] and remove the dependence on the profile different values. With,

Average =  0  = {0, 5, 10}

Minimum  = -1  = {0, 4, 8}

Maximum = +1 = {0, 6, 12} 

So for

with -1 as %Filling pressure

with 1.001 s as Injection time

with 1 as Duration = {0, 6, 12}

If Duration optimum is 'x', just apply the rule from the previous post

for each profile point value = average + delta * x

with delta = (maximum - minimum) /2 = {0, 1, 2}

Cheers,

[p.wozniacki]

Now I'm clear - thanks a lot, I really apppreciate this!

Best,

Piotr