ewbengineering wrote:
How many out there are doing table assessments vs a comprehensive engineering study to get their arc flash labels? It seems that lately we are seeing other firms are offering a table assessment as an option to a study, at obviously a cheaper fee. I have never used the table method, but it appears that there would still need to be some data collection that has to happen. What is everyone's take on this? I can see how as a Facility Manager or Maintenance Manager, if I just wanted labels to be OSHA compliant, that seems to be a cheaper alternative, but as someone who has done this for a while I realize how critical having the proper data is to getting good results and how say an improper installation can take a incident energy number and make it higher than what you would think it should be.
First off I'd say that personally at this point as an end user I'm probably doing the table method about 50% of the time. Reason being that I'm a field service engineer and probably half of my customers don't have an arc flash study if not more. And of the 50% that are doing it I'd say it's about 75% engineering study and 25% table based.
The problem with the table-based method assuming we're talking about 70E tables is that there are a lot of assumptions baked into the tables, specifically with respect to opening times. To be technically 100% compliant with the tables you need to perform a short circuit and coordination study so that you can get your opening times. Since that already requires a full comprehensive engineering study, it's just a couple more mouse clicks away from a full arc flash study. So 90%+ of the engineering effort is already done and if this is your approach to the table-based method, assuming we're talking about either the tables in 70E or the ones in NESC, then this is kind of pointless.
However the reality is that in my situation in particular where there isn't an engineering study done at least 50% of the time, the tables are essentially better than nothing. Based on the prior (2012 and earlier edition) tables a reotrospective study was published that showed that based on about 40 cases of recorded arc flash incidents mostly from Dupont plants when nothing was used, the injury rates were pretty much 100%. Using the table method of that time decreased it to 50%. And following the "IEEE 1584 method" got it down to 0.0% assuming that the PPE was worn as specified by the IEEE 1584 method. There was some second-guessing going on too because I think they recalculated some incidents so show that if PPE was worn following IEEE 1584/ASTM 1959 no injury occurred whereas doing anything less than that went down to 50%. Keep in mind this though. The pre-2015 tables allowed for lowering the PPE based on the idea that less PPE was required depending on the likelihood of an arc flash. This isn't the binary decision of the 2015+ tables where you have to decide if an arc flash is likely based on the task or not. It would allow for PPE level 0 (none required), PPE level 2, or PPE level 4 on the same equipment when obviously PPE level 2 never made any kind of sense. So I haven't gone back and checked the previously mentioned study but it seems likely that most of those 50% injuries are probably from under-dressing based on the 70E tables of the time.
Finally there is the consideration that although the whole basis for the 70E/IEEE 1584/ASTM 1959 "engineering method" is pretty much contrived, it is based on the general concept that it is going to stop a fatality about 90% of the time. It is decidedly not a "no injury" standard because for instance we're wearing the same ATPV PPE from head to toe when clearly for instance the arms and hands are going to be exposed to much higher levels of incident energy. However the practical reality is that so far the success rate is 100%, and the above mentioned study recognizes that there are 1st degree burns for instance in some cases. However the point is that this is about survivability. There has been commentary made in several peer reviewed papers suggesting that if arc resistant PPE is being worn, even if it is under-rated for the condition, since it doesn't sustain a flame, the likelihood is that the area of exposure on the body is less than 25% and thus even under-rated PPE is going to perform nearly as good as properly rated PPE. So as a matter of survival, the chances are pretty good using the current 70E-2015 or later tabular method, whether or not the proper level of PPE is being worn as long as it is arc rated.
In actual practice I'm going to suggest that a hybrid method is the best in the first place in two specific instances. Specifically IEEE 1584 contains multiple methods. The only calculation method that is being referred to most of the time is the empirical method. This method is only claimed to be valid down to 208 V, and the 208 V results are based on a single measurement. All other 208 V tests failed to sustain an arc. Somewhere between around 200 and 300 V, arcs become non-self sustaining. So it can be argued that realistically IEEE 1584 empirical method isn't even valid below around 250-300 V. I would argue that at that point we should look towards a table or "boundary" method in the first place...basically we look at a minimal number of factors such as short circuit current and then assign an incident energy level based on test results which test actual equipment performance such as those done by EPRI that are published in the tables in NESC. Similarly certain types of equipment such as padmount transformers and some meter sockets simply don't "fit" the laboratory test scenarios used in IEEE 1584-2002 so again these should be estimated based on tables rather than IEEE 1584 empirical calculations. In particular meter sockets often exceed IEEE 1584 estimates by way of example.
The second reason to keep the tables is because IEEE 1584 only focuses on calculating how bad an arc flash could be if it occurs. It does not address the likelihood of occurrence. 70E has always left this up to the end user but only spelled it out in the definition of an arc flash, and most people don't read the definitions that closely. It also spelled it out somewhat in the old tables by referring to "H/RC 0" and in the new tables with a new task table that determines if PPE is required or not. The engineering method is pretty much required to develop a similar table but most engineering firms want to run software calculations instead of conducting risk assessment meetings and fail to develop this part of the analysis at all, leaving end users with some pretty ridiculous situations like wearing a 40 cal/cm2 suit to push a start button.
So I would argue that yes tables are being used or at least my company uses them as a contract service provider because often we have no choice in the matter. Granted these are often customers who haven't got a clue. Second I would argue that the tables are not as bad as they are purported to be in the first place. Given the low likelihood of an incident if arc flash is the only reason for doing an engineering study, the improvement may not be worth the money especially for smaller operations. The third argument for table-based methods is simply that you can't honestly do a truly comprehensive study without dong the analysis of at least some portion of it without using tables.
Ultimately looking forward there have been a smattering of both technical and nontechnical papers around a concept that is generally known as a "boundary method" and ultimately this s probably where 70E should go and pretty much where NESC is at on the low voltage side. There are different approaches to it but the general concept is to approach arc flash from a point of view of estimating worst case conditions based on very relaxed information, the kind that say a contract service engineer like me would have access to. I can probably for instance find the transformer feeding a particular piece of equipment and I can probably find and possibly even read the information on the overcurrent protective devices upstream of it. But that's about all that I can locate easily. A boundary value method would take this minimal information and produce a maximum potential incident energy estimate. No comprehensive engineering study needed and furthermore if we say estimate whether the incident energy is no more than a set of discrete values such as say 1.2, 4, 8, 12, and 40 cal/cm2, we can do it with simple lookup tables, not too different from the existing 70E tables.
How would this work? There are a few ways. The first one is that we could for instance do actual testing on actual equipment in a high current lab subjecting the equipment to various real-world scenarios and measuring incident energy of the result. This will result in much lower incident energy levels than the IEEE 1584 empirical calculation estimates and the test results would be linked with a simply parameter to calculate an upper bound on such as short circuit current which often can be found somewhat by assuming that engineering was done properly and looking at the AIC rating of the equipment or by calculating transformer short circuit current which can be calculated from name plate data. A second method is to fix all but a few parameters in the IEEE 1584 empirical equation and then look at the maximum value from the equation or equivalently look at the test data seeking the maximum incident energy with those assumptions. The fixed values are of course things we can find quickly and easily such as the maximum short circuit current. A third method is to look at the upstream overcurrent protective device (if there is one) and recognize that incident energy has a maximum value possible based on the time-current-curve of the device. The only additional consideration would be fault current flowing from other sources such as motors, transformers, and long shielded cables.
The boundary methods have the potential advantage of eliminating the weakness of the existing 70E tables because the need to know short circuit and opening times goes away, or at least might be implicit in the tables. So far the theoretical concepts and ideas have been published in a smattering of papers but none have provided specific and detailed information. There is a commercial company which has published a set of tables. I've looked through the actual underlying calculations. From what I can tell it is pretty good. I can't argue that they don't work but I did have a few questions that went unanswered.