It is currently Tue Dec 07, 2021 4:31 pm



Post new topic Reply to topic
Author Message
ekstra   ara
 Post subject: Arcmaster Critique
PostPosted: Tue Jan 05, 2016 6:59 am 
Plasma Level
User avatar

Joined: Tue Oct 26, 2010 9:08 am
Posts: 2174
Location: North Carolina
Starting a new thread out of the ArcPro criticisms...

http://arcadvisor.com/faq/single-phase-arc

Equations 1, 3, and 4 and the "iterative" approach is identical to Ammerman's approach for DC arcs. A comparison to real world data for DC arcs is given here:

http://www.battcon.com/PapersFinal2012/ ... 0Flash.pdf

I forgot the exact source but some work on the time required for evolution of an arc has been measured and it happens at the nanosecond scale...essentially instantaneously.

The only new twist here is equation 2 which calculates a minimum arc current. This doesn't "look" like any of the DC equations out there that have a similar minimum transition voltage (instead of current) where we go from a low voltage arc where Iarc is strongly dependent on Vs to a point where Varc becomes almost independent of applied voltage and is limited only by system resistance (past the "knee" of the curve). Of course it is certainly possible to express the minimum in terms of current. Still, this one just doesn't look right. I can't attribute it to anything and Arcad doesn't attribute it to anything either.

Equation 9 essentially makes the "Lee assumption", that all arc energy is converted into incident energy. That's obviously not true and is a fundamental problem with all of the DC equations as well as Lee. The only way to calibrate it of course is to use the model on real world test data to derive a fudge factor. That's what the Wilkins model does.

What seems to be missing here is that overall this is still a DC model. It does not do an integration into AC nor is there test data to compare it to anything else so at best, it's an enhanced Lee model on par with the difference between say the original Doan equation for DC (a direct translation of the Lee approach into DC) as compared to the Ammerman model because it takes the exact same logical steps. And sans AC data at best the only test data that can be attributed to it would be the previously mentioned DC test data that was applied to Ammerman's modell.

To turn it into an AC model it needs to be integrated and it needs an arc extinguishment/striking component which is what Wilkins model does. Wilkins compares the results to IEEE 1584 test data and beats the IEEE 1584 empirical model by a significant margin but still within about 10-15% of the same result although it is computationally vastly more complex.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: Arcmaster Critique
PostPosted: Tue Jan 05, 2016 8:34 am 
Sparks Level
User avatar

Joined: Wed Jun 10, 2009 5:00 pm
Posts: 267
Location: Toronto
PaulEngr wrote:
Starting a new thread out of the ArcPro criticisms...

http://arcadvisor.com/faq/single-phase-arc

Equations 1, 3, and 4 and the "iterative" approach is identical to Ammerman's approach for DC arcs. A comparison to real world data for DC arcs is given here:

http://www.battcon.com/PapersFinal2012/ ... 0Flash.pdf


Similar but not quite identical. Equation (1) factors in equivalent source reactance. Reactance does not appear anywhere in Ammerman's equations. Ignoring source reactance when iteratively solving the equations may result in a vary different and wrong answer. Also, if you check the original Ammerman's paper, you'll find out that the author is attributing the equations (2) through (4) to Stokes and Oppenlander. They apply to AC arcs for current decaying from 20kA to 30A.

PaulEngr wrote:
The only new twist here is equation 2 which calculates a minimum arc current. This doesn't "look" like any of the DC equations out there that have a similar minimum transition voltage (instead of current) where we go from a low voltage arc where Iarc is strongly dependent on Vs to a point where Varc becomes almost independent of applied voltage and is limited only by system resistance (past the "knee" of the curve). Of course it is certainly possible to express the minimum in terms of current. Still, this one just doesn't look right. I can't attribute it to anything and Arcad doesn't attribute it to anything either.


Indeed, it is important to watch for minimum arc current and not to apply the iterative approach below the minimum value. I don't agree that arcing voltage is limited only by system resistance. Arcing voltage also depends on gap between conductors, and a combination of system resistance, reactance and voltage, all expressed by the Iarc term.

PaulEngr wrote:
Equation 9 essentially makes the "Lee assumption", that all arc energy is converted into incident energy. That's obviously not true and is a fundamental problem with all of the DC equations as well as Lee. The only way to calibrate it of course is to use the model on real world test data to derive a fudge factor. That's what the Wilkins model does.

What seems to be missing here is that overall this is still a DC model. It does not do an integration into AC nor is there test data to compare it to anything else so at best, it's an enhanced Lee model on par with the difference between say the original Doan equation for DC (a direct translation of the Lee approach into DC) as compared to the Ammerman model because it takes the exact same logical steps. And sans AC data at best the only test data that can be attributed to it would be the previously mentioned DC test data that was applied to Ammerman's modell.


The article reads "This formula assumes the radiant heat transfer. Not all of the arc energy will be transferred as radiant heat as a portion of the energy is transformed into pressure rise and in heat conducted to the electrode material. Therefore, the equation (9) will produce a conservative but safe estimate of incident energy exposure." Wilkins factors show up in equation (10). Also, the Stokes and Oppenlander model discussed in the Ammerman's paper does apply to AC arcs. Quoting Stokes and Oppenlander: "Current and voltage signals have been recorded for arcs burning with exponentially decaying currents from 1,000 to 0.1A, and 50Hz arcs for sinusoidal currents with amplitudes decaying from 20kA to 30A"

PaulEngr wrote:
To turn it into an AC model it needs to be integrated and it needs an arc extinguishment/striking component which is what Wilkins model does. Wilkins compares the results to IEEE 1584 test data and beats the IEEE 1584 empirical model by a significant margin but still within about 10-15% of the same result although it is computationally vastly more complex.


It is in fact an AC model as explained above. The program calculation procedure is well documented and can be found online at http://arcadvisor.com/faq/single-phase-arc or in the program Help. It is also important to note that the ArcPro is proprietary. It does not reveal the actual equations or computational algorithm, making impossible for anyone except the developer to scrutinize the model it is based on and ascertain its validity. You can try ArcMaster before you buy it. It also comes at a fraction of cost of competitor offerings and does not bind you with any "maintenance", annual license renewal fees or other hidden costs.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: Arcmaster Critique
PostPosted: Thu Jan 07, 2016 1:25 pm 
Plasma Level
User avatar

Joined: Tue Oct 26, 2010 9:08 am
Posts: 2174
Location: North Carolina
Quote:
Indeed, it is important to watch for minimum arc current and not to apply the iterative approach below the minimum value. I don't agree that arcing voltage is limited only by system resistance. Arcing voltage also depends on gap between conductors, and a combination of system resistance, reactance and voltage, all expressed by the Iarc term.


This is partly where I have a problem with the model as it stands. I think we can both agree that there is a minimum applied voltage before an arc will occur that is dependent on 3 terms based on the line of research starting roughly with Ayrton and leading up to Stokes & Opplander. It is dependent on the electrode materials, gap, air temperature in the potential arc column, and available current. Two issues arise with modeling this. First, IEEE 1584 test data to date uses only copper electrodes. Second air temperature is something that would be hard to quantify in the first place. Second, it is known that after a few cycles it stabilizes and that it is limited on the upper end by dissociation of the air molecules (turns into plasma) but again, nobody is going to be able to model this on a purely theoretical basis.

In comparison with a DC arc there is no extinguishment. Once an arc ignites, it stays lit. At "low" voltages especially as we approach somewhere around 150-200 V the non-zero ignition of the arc dominates the results. So if we are modeling something in the >10 kV world where the effect is far less than 1% then I'd agree but if this is intended as a general purpose model then I'd have to strongly disagree with the idea that initial arc ignition is not a factor. Granted this grew out a critique of ArcPro whose major users are looking at the world above 10 kV.

Quote:
It is in fact an AC model as explained above. The program calculation procedure is well documented and can be found online at http://arcadvisor.com/faq/single-phase-arc or in the program Help. It is also important to note that the ArcPro is proprietary. It does not reveal the actual equations or computational algorithm, making impossible for anyone except the developer to scrutinize the model it is based on and ascertain its validity. You can try ArcMaster before you buy it. It also comes at a fraction of cost of competitor offerings and does not bind you with any "maintenance", annual license renewal fees or other hidden costs.


It would really help sell your software if you could give a comparison (scatter chart) between actual test data and the output of your software.

Second I can't entirely agree with the statement that ArcPro is "impossible" for anyone except the developer to scrutinize the model. This is not true. Anyone can enter data into the program and get a result. If it is implemented correctly then it should be dterministic and stateless...it should produce the same result for an identical set of inputs regardless of the history of prior inputs. Thus it should be possible to mechanically search the output space and if nothing else develop a "lookup table" that duplicates the software identically. In reality of course it would be much better to curve fit the outputs and derive a set of equations that produces an output with much less storage space. This approach does not need any theoretical basis whatsoever, just as the IEEE 1584 empirical model is purely a curve fit without a theoretical basis for the resulting equations. This approach would be great for developing a commercial or free competitor to ArcPro but has the obvious problem that you stated which is that the resulting equations are simply math. There is no "theory" and no way to scrutinize the "theory" that they are based on.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: Arcmaster Critique
PostPosted: Thu Jan 07, 2016 10:00 pm 
Sparks Level
User avatar

Joined: Wed Jun 10, 2009 5:00 pm
Posts: 267
Location: Toronto
One may argue as well that an arc depends on air pressure, humidity, atmosphere composition, electrode configuration etc. Neither IEEE 1584 nor ArcPro seem to factor the above parameters in the analysis. Hence, in this regard the ArcMaster model is not any worse than the cited calculation solutions.

Since Stokes and Oppenlander V-I characteristics list rms current values with the arc restriking factored in, the Stokes and Oppenlander power equation can be used for incident energy calculations. Indeed, low voltage / low power / long arcs tend to be intermittent yet not necessarily completely extinguished. In such case, circuit protection devices operate slowly or do not operate at all. This type of arcs is considered the most hazardous in regard to fires being ignited from arcing, and relatively safe in regard of skin burns.

Regarding the ArcPro comment, there is no way to find out if it is implemented correctly and if it is based on sound test data and/or theory unless the developer unveils the data. In other words, before you can compare apple to an apple in black box, first you have to pull out the apple from the box. Why would anyone bother developing a lookup table or reverse engineer the ArcPro program if an alternate solution is available, and the problem can be solved by a combination of Stokes and Oppenlander equations derived from the exhaustive study and the R.Wilkins equation developed using the 1584 data for different enclosure sizes. It's up to end user to take ArcMaster with its disclosed calculation procedure, or leave it and embrace ArcPro with its proprietary never published data, no theory an no guidance whatsoever.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: Arcmaster Critique
PostPosted: Sat Jan 09, 2016 7:50 am 
Plasma Level
User avatar

Joined: Tue Oct 26, 2010 9:08 am
Posts: 2174
Location: North Carolina
[quote="arcad"
Regarding the ArcPro comment, there is no way to find out if it is implemented correctly and if it is based on sound test data and/or theory unless the developer unveils the data. In other words, before you can compare apple to an apple in black box, first you have to pull out the apple from the box. Why would anyone bother developing a lookup table or reverse engineer the ArcPro program if an alternate solution is available, and the problem can be solved by a combination of Stokes and Oppenlander equations derived from the exhaustive study and the R.Wilkins equation developed using the 1584 data for different enclosure sizes. It's up to end user to take ArcMaster with its disclosed calculation procedure, or leave it and embrace ArcPro with its proprietary never published data, no theory an no guidance whatsoever.[/quote]

1. OSHA has "blessed" ArcPro. I don't like the approach because they basically compared 1 theoretical approach, 1 empirical approach, and two closed source software programs to each other without reference to test data.
2. If I were a competitor against any of these without a compelling (recognized or tested or something) value, having OSHA "endorsement" is pretty compelling and not just a cheaper price or better documentation.
3. Lacking this, I can always use the lookup table approach to clone it. The biggest value would be incorporating it into other software packages since ArcPro is not available as a software module/library. It may run afoul of some IP issues, but that's a legal question, not a software one.
4. Reverse engineering ArcPro has advantages if the goal is to develop or promote a competing model. It is less expensive for a college grad student for instance to plug numbers into a model or software to get values than to buy IEEE 1584 or the more recent test database that exists. Of course this becomes a model of a model so we're out on a limb and no better than OSHA's treatment but there are real world parallels. The MPEG formats if implemented directly require significant compute horsepower. But recognizing that as long as the output is at least approximately right allowed MPEG layer 3 to be implemented on low end PC's in the early 1990's, something previously only possible with special purpose hardware. ArcPro is very complicated if implemented as described, certainly not as simple as ArcMaster.
4. If we can duplicate results with a documented and lower cost software (your argument) and without real world test data and if whether we agree with the politics of it or not the accepted standard is ArcPro, why not compare the two directly?

I should say my motivation here is not to defend ArcPro, far from it. I'd like to see ArcMaster or something similar displace it. Privette's software (aka Duke Heat Flux) has ArcMaster beat on price (free) but compared to the gold standard, it was not accepted. I'm just asking to subject ArcMaster to the same scrutiny. At 10-15kv we have test data and we have all 4 models. If ArcMaster holds up similar at 10-15 kV it makes a compelling case above 15 kV. There are also some EPRI equations and data for mid-range long arc (6-12") cases for comparison, too, since Stokes & Opplander should hold until the arc is affected by convection and phenomena at the ends.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: Arcmaster Critique
PostPosted: Tue Jan 12, 2016 7:38 pm 
Sparks Level
User avatar

Joined: Wed Jun 10, 2009 5:00 pm
Posts: 267
Location: Toronto
PaulEngr wrote:
It would really help sell your software if you could give a comparison (scatter chart) between actual test data and the output of your software.


Check please Table 1 online at http://arcadvisor.com/faq/single-phase-arc for the comparison of arcing currents measured by two laboratories and the current values calculated using ArcMaster. Note that both in theory and practice the arcing currents are dependent on system power factor when even a small change in the power factor may result in a significant change of arcing current. I don't really recall that ArcPro is factoring system power factor in the analysis.


Top
 Profile Send private message  
Reply with quote  
 Post subject: Re: Arcmaster Critique
PostPosted: Wed Jan 13, 2016 2:19 pm 
Plasma Level
User avatar

Joined: Tue Oct 26, 2010 9:08 am
Posts: 2174
Location: North Carolina
arcad wrote:
PaulEngr wrote:
It would really help sell your software if you could give a comparison (scatter chart) between actual test data and the output of your software.


Check please Table 1 online at http://arcadvisor.com/faq/single-phase-arc for the comparison of arcing currents measured by two laboratories and the current values calculated using ArcMaster. Note that both in theory and practice the arcing currents are dependent on system power factor when even a small change in the power factor may result in a significant change of arcing current. I don't really recall that ArcPro is factoring system power factor in the analysis.


We're going from the world above 15 kV to talking about 120-263 V. There is a huge difference here. Below 250 V, it is hard to get an arc to be self-sustaining and arcs can even extinguish and then restrike later on. A similar small (but not huge) adjustment occurs in Ammerman's DC arcing model which is the model most similar to yours as compared to Lee which assumes a fixed value that I_arcing = I_bolted*0.5.

But this doesn't point to accuracy for incident energy. Below 600 V despite obvious modeling errors not only in terms of current but the fact that voltage is not really directly a factor as voltage increases, Lee is known to have reasonably good agreement with incident energy test data despite obvious modelling errors. All that it shows is good agreement with an internal model parameter (arcing current) without showing good agreement overall (incident energy).

What I'd like to see is ArcMaster incident energy vs. measured incident energy, and it would be even better to show it alongside ArcPro or IEEE 1584. If there is good or better agreement, then it's pretty easy to justify. But with only current data this doesn't best ArcPro (OSHA "approved") or IEEE 1584 (good agreement with incident energy test data).


Top
 Profile Send private message  
Reply with quote  
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 7 posts ] 

All times are UTC - 7 hours


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Jump to:  
© 2019 Arcflash Forum / Brainfiller, Inc. | P.O. Box 12024 | Scottsdale, AZ 85267 USA | 800-874-8883