While performing work for a customer I discovered numerous issues with a recent arc flash study they had performed by another contractor. I've been asked to provide code references that support my belief as to why things that were done are in deed wrong. Below are some of the more critical issues discovered.

1) Transformers were labeled but list their shock hazard as the highest value from the secondary side (where they analyzed it) instead of the highest voltage present which is that of the primary side. [My recommendation is to remove these labels based on the belief that no one will access them while energized.]

2) Devices are incorrectly labeled - In some cases the system was modeled such that the main had a label and also all feeder breakers had a label unique to them. The model was setup so that the main acted as the OCPD for the feeder breakers. On other panels they modeled it such that the main was not the OCPD. In my experience the customer should make the call as to which way to analyze panels and then all equipment in the plant should be analyzed the same way. Another issue is that labels that were found to have been for field equipment (control panels, disconnects, etc.) were incorrectly placed on the feeder switches at the upstream distribution panel. This was proven by matching up the equipment ID's on the labels to the model.

3) In multiple cases they did not trace out equipment to the appropriate utility feed. Instead, in their model they used a "generic" 500kVA transformer. With little effort I was able to trace one to a 2500kVA transformer and another to a 1500kVA transformer.

4) The 12.47kv to 480V step-down transformers were not configured correctly in the software after reviewing their nameplates. Incorrect impedance values and temperature rises were used.

5) IE values do not match up throughout the project. In one (of many) cases a distribution panel is labeled with an IE of 3.9. The report they provided my customer indicates its IE is 1.6. When I analyzed the model I found it to be 39.5.

6) In the review of one of their power systems approximately 80% of the circuit breakers with trip settings were modeled with settings that do not match what I found in the field. Their reports also did not recommend setting them to what was modeled so it appears the model simply does not match the power system.

7) In many cases labels from previous studies were left. Some distribution panels have labels from three different studies. In one case the value dropped from 19.2 cal down to 1.11 cal. In another case there are two labels with different IE values from the most current study. The two labels have dates just a few months a part.

8) Of the approximately 1,000 points where the AFH should have been calculated only one motor was modeled.

I am not aware of specific codes that justify my argument that the above items are bad. I believe my supporting standard is plain and simple...OSHA "Employers are to provide a workplace free of known hazards".

I'm curious what everyone's thoughts are and if you are aware of any specific codes that pertain to any of these items.

I can't tell from your description if all of your issues are valid or not since I'm not there but there is no standard for a lot of arc flash engineering studies other than IEEE 1584.1 which only came out recently. All that OSHA provides is the generic "general duty" clause which simply does not set standards for how to do an arc flash engineering study. So the only issue to be raised here is engineering malfeasance which is extremely hard to prove since there is no standard for how to do things. Our local (NC) board often goes after surveyors because they didn't follow standards but lacking such standards for the most part for engineering virtually all of the citations against engineers are for practicing without a license.

So playing devil's advocate and also offering some counterpoints to your arguments...

Taking your first item with a shock hazard listed only for the low side, where is the oil sample port located? Most of the time on pad mounts (utility style) it's on the low side but once in a while I've seen it on the high side. That is a perfect example of a reason to open a door while energized. Similarly if infrared scans are done that's another reason. Another reason...is the high side using elbow connectors? If so then the high side wiring is 100% enclosed inside shielded connections and the connectors are either load break or non-load break but either way, they are actually miniature disconnects, especially if there is a parking stand. Finally, is there a practice in place to always de-energize before opening the inner/second door for the high side? Since it's medium voltage some plants elect not to work medium voltage (and have no one trained) while others do. So if that's the case, the labels may only apply to the areas in which their electricians are qualified. I have several customers like that. In either case I'd challenge you to demonstrate where and how a shock hazard would exist on the high side but as far as whether or not to label the low side it depends on work practices and rules. I doubt you'd find anyone trying to disconnect anything while hot unless it falls under 1910.269 or a very aggressive interpretation of Subchapter S which you typically see in hospitals.

Similarly with item #2 this reads to me at least like engineer A has opinion A and B has opinion B about how to do something. There is an ongoing argument about when and where arc flash can "jump" over a main OCPD with no firm guidance anywhere. So picking one or the other can indeed be arbitrary based on the opinion of the guy who did the work.

In terms of item 2B (mislabelling) as well as having multiple labels (item 7) this depends on who put the labels on. If the engineering company did it then shame on them. But if for instance the owner's own electricians did it, first off I've been handed a pack of labels that someone else did and often struggled to figure out what their intended label scheme actually is. And electricians and mechanics are generally taught not to remove/disturb existing labels because those are also important and/or might be necessary from a history point of view and you don't know what can and can't (or should/shouldn't) be destroyed, and when it is important to make sure that you do remove old labels. It's a lack of training in this case. So the engineering firm might not even be guilty of that one. Did the engineering firm do the training too? Did they do any testing or verification that the training "took"?

In terms of motors, this gets into estimating. I don't know what set of rules you learned by the rules I learned for motors are to ignore anything less than 25-50 HP. From around 50 to 250 HP add them all together and represent them as a single motor. At 250 HP and over, represent them by individual motors. This is based on error analysis. My argument as an end user is then you can't really figure out what anything is and it takes to little time to add the motors individually that other than license limits, why not just add them all because then the arc flash software output becomes a single line for the plant rather than relying on a bunch of single lines that rarely if ever get updated and again the old ones never get tossed.

The rest of your issues point to model errors. Often engineering firms are pushed to put out results without actually doing a good job of tracking everything down, and not even doing an adequate job of sampling and checking previous results. Particularly in large plants most of the time the models are simply rife with errors. I've seen terrible work done even by very large international firms (BE&K/KBR) filled with all kinds of errors. All that I can say is that I'm pretty sure that especially with very large plants (over 1000 buses) this stuff gradually improves in terms of accuracy over time. The big problem that I see with it is that in the quest to get costs down virtually all engineering firms long ago got rid of checkers. There are still blanks on the title blocks for draftsman, checker, and approver, but either the same name is used on all 3 or there is only one signature on the drawing. And if you hand a pile of drawings and/or labels to the owner and ask them to check it over and/or approve it, they don't have the resources either so most of the time they simply rubber stamp it so that nobody really did a detailed inspection like you did.

Finally often pricing is negotiated. Basically the owner has a dollar figure they can spend so you negotiate a price based on how far you go in collecting data. This may mean that you accept old data and/or "simple" it and do some updates based on that. You should see statements somewhere either in the report or the quote that mentions that the source of the data might be based on data provided by the customer (old reports) for instance. This isn't the ideal world but recognizing that with 1000+ buses the price for the study will be 6 digits and trying to provide as good of a service as possible while keeping within bounds of what the customer can pay. This again isn't the ideal world and obviously careful wording by the engineering firm steers clear of the issues, particularly when they know that they exist, but also trying to achieve a balance between costs and quality of the work product.

Look at it this way...if you can successfully convince the customer of the issues you found and convince them that your detailed way of doing things is better, you can make a pretty good case for throwing the competitor out, virtually guaranteeing you repeat business in the future. If you feel that you have an ethical issue with what the previous firms did (and you probably should) then if you have a PE you are required to file the complaint with the state board. Just as I said though you have to stick with the things that are well settled and documented because otherwise its just one engineer's opinion against another's. Or from a business point of view recognize that work quality is a problem everywhere, use it as a sales pitch, and move on. I just tell customers about my findings in a nice and/or high level point of view and some of the issues with it, and move on. Let them make their own decisions. Engineers are not gods nor responsible for saving humanity from itself. We have a job to do, and it's a matter of pride, of ethics, and ultimately the way to get repeat business to do everything you can to do a quality job every time.

So step back a minute and blow off a little steam. Recognize that this is an opportunity to correct some things and make it better. Don't "go negative" because aside from a little advertising you can come off sounding like a negative or sneaky person that might not be trustworthy if you don't stick to the facts but also put the past in the past. Take it from someone who was always a pure black-and-white person who recognizes that we live in a world of grey.

PaulEngr, thanks for your thoughts on this. They're much appreciated.

Just to follow up on some of your comments/questions...

For item 2 I agree it comes down to, hopefully the owner, deciding how to treat the main switch at a panel. From what I've learned they were never asked how they wanted this handled. That is unfortunate but my issue is that they were not consistent throughout the plant. Some panels have low values when they used the main as the OCPD while others are high because they did not use it as the OCPD. The amount of inconsistencies and errors I've found from their work makes me not trust a single label they applied.

As far as the issues with labeling I confirmed that the contractor collected the data, modeled/analyzed, and installed all labels themselves. I initially feared that labels were delivered to the customer and they had their own personnel apply them. Since it was the contractor that did the modeling there is no reason they should consistently be applied incorrectly. Now MANY devices are not labeled at all while MANY distribution panels have multiple incorrect labels creating confusion, and in my opinion, an unsafe work environment.

In regards to the motors I typically model any motor individually if it is 50HP or greater. Additionally, if there is a grouping of motors totaling 50HP or more fed from a single panel I model them as a grouped motor load. As much as possible I model all motors individually as it is connected to the system. This is the only way to get accurate motor contribution back onto the electrical system which is important when analyzing the system for available fault current.

Finally, when it comes to the quality of data collecting versus funding available on the job I feel that identifying the arc flash hazard is a very important thing. If a customer cannot provide funding to allow the accurate modeling of their entire system then they should try to take on small areas of the plant and do it in phases. The company that performed the study has done the study once and now has completed two 5-year reevaluations. If the system was accurately verified/modeled years ago then the reevaluations should be as easy as verifying loads at all panels and identifying where changes/removals/additions have occurred.

My response is based purely from experience being handed such a mess in the past. At first it infuriated me. Then I calmed down and thought about the rationality of why it ended up like it did and what I can realistically and practically do about it. During the next "go around" in terms of arc flash studies we actually reduced the costs and had very specific directives for the contract engineering firm which helped improve things greatly because we were all on the same page, AND I sent and intern around with electricians as part of his training to do a lot of the leg work ahead of time so that the engineering firm was most just validating our model updates. More of what a 5 year update should look like instead of a 5 year "total redo".



The single lines in the largest facility I've handled (over 2000 buses) are in an awful state. Every time a project was done, the drawing vault was basically a dumping ground for work prints and as-builts. So you could sort of make out every single stage of every project with the layers upon layers of conflicting drawings, when drawings actually made it to engineering. And we all know how that works in the real world. So the reality is that the power system analysis "drawings" are actually the most accurate drawings of the power system available. So we modeled each motor individually for that reason alone. But getting back to your point, up to a point simply adding up all the motors (basically, inductors) in say an MCC gives you the X' and X'' you need, up to a point. That's why the standards recommend adding motors together like I described. The error that it may induce is very small. Try it some time and you'll see it saves a lot of time, but it does distort the power system analysis drawings if your intent is to use them as single lines as I described. It's hard to find that "145 HP" motor in the plant.



Actually it's even easier. Other than reviewing changes in the 5 year window it should be as simple as reviewing a lot of "soft" factors such as quizzing personnel to evaluate their knowledge of arc flash hazards and doing a job inspection to verify how the work procedures as they are followed, reviewing any safety inspections being done for energized work, energized work permits, etc. and doing a "random" sampling of existing areas to verify that things are not changing. This would be for an arc flash risk assessment as opposed to an arc flash hazard assessment which would be entirely devoted to looking at verifying that the power system model and the actual plant are one and the same. Obviously the cost of doing so is as you suggested very modest assuming the initial arc flash hazard assessment was done thoroughly and correctly.