Wanting to Test Vacuums

[Absolute Rainbow]

The weaknesses I can see are as follows:

• No procedure and measurements outlined to ensure consistent initialisation
• No discussion of the distribution and placement of dust within pile — “spread evenly” is extremely vague and far more needs to be considered
• No discussion of the properties of the dust materials and learned explanation of why they’re representative of an environment the machines were designed for
• No discussion of the important properties of the carpet
• No quantification of the cleaning speed and understanding why this is important, including overlap properties
• No discussion of how to ensure accuracy of what’s measured and priming system to account for mass losses within the machine (dust sticking etc. and not being weighed)
• No recognition, appreciation, or understand the statistical nature of particle removal from first order systems and how this is to be captured and correctly interpreted
• No evidence of data reproducibility in any testing methodology, given all the above

Most people don’t have PhDs in experimentalism, so this is not surprising. Much of this is covered in my videos.

Even if you did all the above, without a laboratory and the necessary training, the results still have to be taken with a pinch of salt owing to lack of rigor that not using a laboratory brings.

@cheesewonton

 

[Vacuum Facts]

The trend of the vacuum's removal quantity of (remaining) dirt per pass is not stochastic.

Agreed. That's not what it means though. See all evidence and explanations above as this is all fully addressed.

Now, I suppose technically if you want to analyse whether or not each individual particle of dirt is removed or not, I suppose you could stretch that into calling 'that' stochastic.

Individual particles are removed and some aren't. This is exactly what is meant by stochastic above. It is precisely mathematically modelled and empirically confirmed. There's literally no debate here and this is only revealing tremendous ignorance of well understood and solid science. Anyone who doubts this simply needs to go away, learn it all, and understand it. It has all been provided. If anyone here, of all places, thinks they know better, they'll carry no respect until they publish their findings like the actual scientists did and explain why they're contradicting them.

 

[herbicide]

Individual particles are removed and some aren't. This is exactly what is meant by stochastic above. It is precisely mathematically modelled and empirically confirmed. There's literally no debate here and this is only revealing tremendous ignorance of well understood and solid science. Anyone who doubts this simply needs to go away, learn it all, and understand it. It has all been provided. If anyone here, of all places, thinks they know better, they'll carry no respect until they publish their findings like the actual scientists did and explain why they're contradicting them.

I challenge you to reconsider. If a particle is not picked on the first pass, is it:

1. More probable to be picked up on subsequent passes
2. Less probable to be picked up on subsequent passes
3. Same probability to be picked up on subsequent passes

As we’ve already agreed upon the answer is “2,” less probable. This means by definition the behavior is not stochastic; ie randomly determined.

 

[Vacuum Facts]

You can determine this mathematically from the verified models provided above if you truly care. I recommend understanding it all if that's your goal. You clearly currently don't understand what stochastic means and the models provided for you.

 

[herbicide]

You can determine this mathematically from the verified models provided above if you truly care. I recommend understanding it all if that's your goal. You clearly currently don't understand what stochastic means and the models provided for you.

“Verified” by whom?

What “models?”

 

[centralsweeper63]

The weaknesses I can see are as follows:

• No procedure and measurements outlined to ensure consistent initialisation
• No discussion of the distribution and placement of dust within pile — “spread evenly” is extremely vague and far more needs to be considered
• No discussion of the properties of the dust materials and learned explanation of why they’re representative of an environment the machines were designed for
• No discussion of the important properties of the carpet
• No quantification of the cleaning speed and understanding why this is important, including overlap properties
• No discussion of how to ensure accuracy of what’s measured and priming system to account for mass losses within the machine (dust sticking etc. and not being weighed)
• No recognition, appreciation, or understand the statistical nature of particle removal from first order systems and how this is to be captured and correctly interpreted
• No evidence of data reproducibility in any testing methodology, given all the above

Most people don’t have PhDs in experimentalism, so this is not surprising. Much of this is covered in my videos.

Even if you did all the above, without a laboratory and the necessary training, the results still have to be taken with a pinch of salt owing to lack of rigor that not using a laboratory brings.

Here you stated the problems.

To distribute dust you can use a flour sifter ( the type with a pull trigger that is used in dusting a cake in icing sugar.) and count how many times you pull the trigger. This allows you to pull the trigger and release to evenly cover the rug. 50 grams of dirt is about right for a 4x4 foot rug. In a home you usually encounter dust, grit, fluff and larger debris. Dust and grit is replicated by a mix of equal parts flour, baking soda and fine sand. Fluff is replicated by kpok, which is spread by holding a large clump and holding it about a centimetre above the carpet and letting it stick how it wishes. larger debris can be replicated by holding about 100 grams of clean cat litter in your hand and holding MOST but NOT all and throwing it randomly over the rug. I haven't measured how fast I move the vacuum across the rug. You could take the size of the rug and time how long it takes to do 100 passes figure that out and then calculate the average speed based on that. I vacuum in a w being straight forward diagonal back and repeat. Overlapping 50% is my standard. If I understand your second to last point correctly, science and theory has no effect on a practical test other than the hypotheses. This is why we test, to avoid unknowns and practically prove them correct or incorrect. To account for particle sticking etc you simply weigh the whole machine before and after. weighing the carpet proves what goes in and out. Hence the record keeping. The important properties of the carpet are as follows: two pile, This is tricky for vacuums as there is a loose tall pile of handle then also a very denso short pile bellow. This makes cleaning extremely difficult. Wool loop pile, wool loop pile carpet is great for creating traps and block as it is very densely made. Dust in the centre of a loop is difficult to clean!. ZSolution dyed nylon is great to represent a typical average carpet. Mohawk carpet is great for representing thedeeper piles that trap dirt. an area the size of a bottle cap can hold up to roughly 2-3 teaspoons of dust without looking terrible. Hence it is terrible to fully clean and know it is clean.

Here I responded with corrections to my method, then you quote your earlier reply that needs updating.

 

[cheesewonton]

I challenge you to reconsider. If a particle is not picked on the first pass, is it:

1. More probable to be picked up on subsequent passes
2. Less probable to be picked up on subsequent passes
3. Same probability to be picked up on subsequent passes

As we’ve already agreed upon the answer is “2,” less probable. This means by definition the behavior is not stochastic; ie randomly determined.

Again, wrong. The rate of removal per pass is not random if the test procedure is correctly accomplished. The rate of removal should be a line of decreasing slope. That is not what you get with a random distribution.

 

[cheesewonton]

The trend of the vacuum's removal quantity of (remaining) dirt per pass is not stochastic. If it was stochastic each pass would randomly remove more, less, or the same amount of dirt, instead of a predictable regression. Variance doesn't make it stochastic.

It cannot be both stochastic, and the below:



Now, I suppose technically if you want to analyse whether or not each individual particle of dirt is removed or not, I suppose you could stretch that into calling 'that' stochastic. But even that is a stretch, because we already know (ahem) that if it not picked up on the first pass, its chances to get picked up on each proceeding pass logarithmically regress.

Stochastic means random. There is nothing random about the rate of dust removal per pass. It is a negatively sloping curve. The first pass gets the most dirt, and each subsequent pass removes less. Nothing even a tiny bit random about that.

 

[herbicide]

Again, wrong. The rate of removal per pass is not random if the test procedure is correctly accomplished. The rate of removal should be a line of decreasing slope. That is not what you get with a random distribution.

We're saying the same thing (its not stochastic)

 

[Vacuum Facts]

Stochastic means random. There is nothing random about the rate of dust removal per pass. It is a negatively sloping curve. The first pass gets the most dirt, and each subsequent pass removes less. Nothing even a tiny bit random about that.

That's right. All this is discussed above and links to the papers provided. I recommend you read to understand seeing as you're clearly not listening to me. It's amazing how you just aren't able to listen and read what's provided for you. It's so obvious where you're confused as well, but there's just no helping someone not interested in understanding what is being said.

 

[Vacuum Facts]

Here you stated the problems.
Here I responded with corrections to my method, then you quote your earlier reply that needs updating.

Unfortunately, the reality is, I don't have time to play personal tutor. All the information required to do things well is available. It's down to you at this stage to do your own research, understand things, and make your best judgement. You may want to download the standards and study them. You might want to understand the background science. It's going to be tough without training. If you get it right, your data will be accepted, if not, it won't.

 

[centralsweeper63]

I will try my best to refine my testing methods. I am still in middle school/intermediate school though so I appologise for any sloppiness in my testing.

 

[huskyvacs]

If you want to know how to do airflow tests the right way, VacLab is who you study. He knows what he is doing because he is a real engineer .

 

[Vacuum Facts]

If you want to know how to do airflow tests the right way, VacLab is who you study. He knows what he is doing because he is a real engineer .

This has been widely debunked as a meaningful metric. Using vane anemometers in the way shown in the video is a hallmark of total lack of understanding of basic fluid dynamics and how these instruments work, as fully discussed here. Using a configuration like this is fraught with errors, as also discussed in that thread. Furthermore, this measurement is wholly irrelevant anyway and isn't representative of what needs to be considered, as discussed here and the associated links. It's pretty bad this kind of nonsense is still being linked to and reveals a lot of about the trustworthyness of the sources. Anyone who genuinely understands the science, and desires to, will eventually realise this from the material available now. Fortunately, most of the silent lurkers in this place aren't netted by the primitive tribalism, bad faith, and aggressive ignorance on display at times that otherwise sours a fairly decent forum resource.

 

[Hatsuwr]

If you want to know how to do airflow tests the right way, VacLab is who you study. He knows what he is doing because he is a real engineer .

I like that his method allows testing the interface between the cleaning head and surface being cleaned, instead of just the airflow and suction of the motor minus plumbing losses. I saw on another video that he taped the head to the box to get peak values, but I hope that he does unsealed tests too.

A few thoughts for improvement:

1. Instead of having two different opening sizes, have one side carpeted and make a removable insert for when a different opening size is needed.
2. Make sensor two holes on each side. One for the vacuum gauge, and one for the anemometer. Have a few different sized inserts for the anemometer hole to simulate different levels of restriction
3. It would be interesting to see the effect that turbulence and uneven airflow have on the airflow numbers. I wouldn't expect there is much impact with his setup, but having a hole in the center brace and using the side opposite of the head for the airflow measurements could help if there is any issue.

Still, even without those changes, this is a perfectly valid comparative method for a few aspects of performance.

 

[cheesewonton]

I like that his method allows testing the interface between the cleaning head and surface being cleaned, instead of just the airflow and suction of the motor minus plumbing losses. I saw on another video that he taped the head to the box to get peak values, but I hope that he does unsealed tests too.

A few thoughts for improvement:

1. Instead of having two different opening sizes, have one side carpeted and make a removable insert for when a different opening size is needed.
2. Make sensor two holes on each side. One for the vacuum gauge, and one for the anemometer. Have a few different sized inserts for the anemometer hole to simulate different levels of restriction
3. It would be interesting to see the effect that turbulence and uneven airflow have on the airflow numbers. I wouldn't expect there is much impact with his setup, but having a hole in the center brace and using the side opposite of the head for the airflow measurements could help if there is any issue.

Still, even without those changes, this is a perfectly valid comparative method for a few aspects of performance.

His method overestimates airflow at the nozzle

 

[Hatsuwr]

His method overestimates airflow at the nozzle

How so?