Why Closed Ring or Line, Multiple Bus, Main Switchboard Designs Need Arc Flash Detection Trips

Introduction:  There seems to be a problem in the Far East.  I’m seeing too many ships with closed bus designs that ignore DP guidelines and lessons from the last 20 years.  It’s partially because big yards can bully class and choose poor local DP enforcers, and partially because the local vendor branches haven’t had to work to DP standards.  They are the low cost leaders and produce most of the world’s ships, but when I get a DP design from there, I tend to see repeated design problems.  Closed bus, redundant DP, electrical designs demand much more than main class, but main class is essentially all that is being delivered in some projects and by some vendors.  This is unacceptable, as we are allowed no single points of failure in DP, but main class doesn’t care.  This means considerable extra protection is needed for a closed bus DP design.  I’m going to look at one representative example – optical arc detection protection.

Background:  I recently had a meeting about all the outstanding items on my DP FMEA for a project.  I’m the bad guy because I am enforcing the rules and protecting the future crew and operations.  I’ve actually done multiple reviews and revisions to help the project to completion, but the shipyard and vendor are still confused, because they are not accustomed to DP rules, despite having done many DP projects, because they aren’t used to rule enforcement.  Finally meeting about the problems, the shipyard couldn’t escape their own stuff and quickly manned-up, but the electrical vendor swore up and down that his design met requirements, despite all the problems that I had identified from his own documents and the clear indication that only main class requirements were considered.  There were so many problems that I could swamp them with a listing, and they had been playing whack-a-mole with comments instead of understanding their need to think through fundamental problems.  So, I focused on one problem that everyone should understand – the lack of proper arc detection as a representative design fault.  They did not understand and asked me for references.  I’ve worked on lots of closed bus projects over the years, all with a clearly defined need for arc detection protection in their design documents, so “No problem,” I foolishly replied.

No Rule:  Late that night, I tried to find easy and decisive rule and guideline references and found almost nothing.  If the project was ABS, the only reference to arc detection is agnostic and for enhanced electrical class (Sec 8 3.1.2.ii, arc detection, coated bus bars, or something else - “please, just class with us, we can be reasonable”), but at least, they directly reference MTS & IMCA standards in their requirements.  They would be slightly more insistent if it was hybrid (Pt 6, Ch 2, Sec 24, 24.27.6).  If it was DNV, the class rules don’t discuss arc detection or reference MTS or IMCA, but their DP design recommended practice (DNV-RP-E306 8.7.2) does, because it is derived from MTS.  So, on their own, class rules give me little beyond “no single points of failure.”  That is good enough in the end, but not useful when dealing with a vendor that made so many basic mistakes that argument from first principles would become a lengthy training session in electrical protection fundamentals.

Few Guidelines?  To avoid that, I started into the industry DP guidelines and started with MTS because it is used by both ABS & DNV.  I was tired, failed at basic reading, and found little (lesson: sleepy = dumb).  So, the next day, all I could give the vendor and yard is TechOp D-05’s requirement for DP FMEAs to review arc detection for closed bus designs but not open bus (item 327 in the table), and IMCA M206 3.11.8 “over current protection is not sufficiently selective to prevent WCFDI being exceeded but it should prevent a fire if the optical system failed for any reason.”  Neither reference is perfect, but they reflect the redundancy reality in combination.  Because I am a sporting fellow, I noted the weakness on the class side, and the vendor jumped on that in the meeting.  I then explained that when you have a low and high requirement in the same standard, you need to use the highest applicable requirement, rather than the lowest.

Back to Guidelines:  The problem that I had with the guidelines is that they state that there are problems with directional overcurrent zone protection, but don’t explicitly require arc detection.  Neither the MTS DP design guidelines nor IMCA M206 put down hard stops, even though all successful closed bus, multiple bus projects do.  I can’t accept a lack of arc protection, when overcurrent will not reliably avoid single point failures, but the guidelines stick to gentle hints.  For example, IMCA M247 notes arc detection is significant but says nothing more, and MTS DP design 11.7.4 says “Differential protection schemes can have problems with high levels of through-fault current. That is current passing through a healthy zone on its way to a fault in some other zone. There have been problems with healthy zones tripping causing failure effects exceeding WCFDI. It is for this reason that some designers favour arc protection for this application.”  That is a very wishy-washy way to note that differential overcurrent isn’t ideal, allows single point failures, and effective arc protection is required where redundancy depends on fast and reliable selectivity.  Similarly, MTS DP design 11.7.2 and IMCA M206 3.11.8 indicate the known serious problems with using just directional overcurrent, but fail to then conclude that single point failures are bad, so put in arc detection.  Arc protection is just icing for open bus (unnecessary), and non-ring, closed bus, 2 split might be able to do without it by tripping the bus ties early (possibly unnecessary), but multiple bus, closed bus, line or ring systems need arc protection for effective fault discrimination, and that’s not directly stated in the guidelines.  We have made the mistake of assuming people know.  For example, only one of my articles has even mentioned arc detection before this.  It’s known by people who have worked in the industry for years, and probably understood in the main office of the vendor, but the vendor subsidiary doing all the work didn’t know.  

Clarification:  I reached out to the fellow who wrote the electrical part of the MTS and IMCA guidelines, and asked him why the guidelines weren’t more explicit about arc detection.  He agreed that modern optical arc detection had proven itself invaluable, but explained that it was less prominent in the guidance because there were problems with pressure based arc detection.  He was right, the pressure detection is slower and prone to common faults, as the air pressure wave can be transmitted to other parts of the switchboard and blackout more than one bus.  There have been related DP incidents.  So, optical arc detection is essential for some designs, but not pushed as hard as it should be in the guidelines, due to problems with pressure-based arc detectors.  Air pressure arc detectors are problematic, but optical arc flash detectors are invaluable for selectivity and speed in multiple closed bus systems.

Basic Principles:  So, why is optical arc detection protection better and faster than differential overcurrent protection?  Obviously, light is faster than electricity, but it’s the complication of multiple contributing parties that really slows things down.  In your mind, replace all the switchboard protection relays with a person doing that job.  In the most straight forward optical design, one man waits for the light showing flash detection in the bus bar area and then presses the trip button to trip all the breakers and isolate that electrical bus.  In the differential overcurrent design, a man at each breaker sees a light indicating overcurrent towards the bus and then calls up all the other men to see if they saw the same thing and agree to trip.  One man acting directly on a clear indication is obviously faster than the deliberation of the meeting, and the meeting can be thrown off by one weak or slow breaker.  With little offshore breaker maintenance, the chance of failure of the more complex scheme, or loss of coordination between zones increases.  The more complex design, the more deliberation required, the greater the chance of faults, and the longer until protection acts, but we need fast, effective protection to prevent single point failures.  Not all bus faults cause a detectable flash though, especially with coated buses or other obstacles to detection, so we want both forms of protection, with the slower and less reliable differential overcurrent backing up the optical arc flash protection.  If it were a race, then we would bet on optical arc detection to win and on differential overcurrent for second place (triggers in milliseconds vs. centiseconds).  Arc detection limits damage and disruption of the power systems, while differential overcurrent will eventually get there.  There is a much better chance of surviving an electrical fault when it is quickly eliminated.

Complication:  While the advantage is plain in direct acting designs, some switchboards have multiple compartments for each bus and need so many detectors (one for each compartment) that each protection relay only covers its section, and a bus spans several sections.  In that case, one man sees the flash signal and presses the trip button and sends a signal to the other men to press theirs.  This is slightly slower.  Even worse are the networked designs with what is essentially a party line for gossip (nice to have non-vital information) and control being shared with emergency trip signals.  The guy who sees the light needs to clear the line and send the trip command to the other guys who have delayed responses because they are distracted by other work.  Some of this is made up for by increased network speed and processor power, but designers like bloatware, so it is a race that safe operation can lose.  If the party line goes down (network failure), then neither the arc flash detection nor differential overprotection will work.  So, there needs to be backup protection (trip lines) or the network needs be almost bulletproof (GOOSE isn’t, it’s just more protected).  So, bad design can negate both optical arc flash and differential overcurrent protection.  Generally speaking, a protection relay that detects an arc flash in the bus bar should directly trigger the bus isolation.  There are means of increasing security by trading speed for more time to confirm signals.  Essentially, you are trading different types of safety margin, too slow and you may as well not have protection, but unconfirmed lets a single electronic fault trip a bus. 

Conclusion:  This is just an introduction.  A well designed optical arc detection system is essential for redundant DP, closed/multiple bus protection due to its better speed and selectivity.  Differential overcurrent or zone protection is a backup with slightly different methodology.  Having fallback protections supports DP redundancy.  This is hinted at in the guidelines, but not explicit.  This may not appear to be covered in the rules, until you understand how the protection systems work and fail, and compare that with the requirement for no single point of failure.  None of this is new to people who have worked on multiple offshore newbuild or refit projects, but it has somehow been lost as global production has shifted, and needs relearned.  Poor design can reduce the advantage of optical arc detection, and bad design and maintenance can remove the effectiveness of both arc detection and differential overcurrent protection.  Now you are warned, and can help keep the vendors honest.