Sunday, October 23, 2016

Engine Room Fire Suppression System

Earlier this month I mentioned that I was installing an automatic fire suppression system in Vector's engine room. Today's post is entirely about the motivation, selection, design, and installation of this system. If you have no interest in such arcane technical matters, feel free to skip it; I will resume our normal travelogue in my next post.

Let me start by saying that it's really inexcusable that a boat such as Vector, designed for open ocean passages and constructed in the 21st century, was not equipped with such a system right from the factory. Almost all recreational vessels of this size today are so equipped by their builders, and ABYC guidelines mandate them. It's particularly baffling considering she was built to Lloyd's shipbuilding standards, and even included a high-volume water pump suitable for manual firefighting (although the hose connection and fire hose were never installed).

For whatever reason, the original owner, who also finished the construction and outfitting of the vessel after the original builder went bankrupt, did not see fit to install such a system, and the boat has lacked one until this month. Our on-board firefighting capabilities have heretofore been limited to a number of hand-held dry-chemical extinguishers, and a garden hose attached to the seawater washdown pump.

A fishing vessel burns in Maine. USCG photo.

Installing such a system has been on my to-do list since the day we took ownership of the boat. That it has taken over three years to accomplish speaks to the complexity and difficulty of the task as much as it does to the general triage of boat projects and their impact on our daily life. Planning for the project kicked into much higher gear after our friends aboard Blossom experienced an engine room fire in the Bahamas; you can see the engine room video of the entire event here. This same event prompted us to install a video camera in the engine room.

Most small-vessel automatic engine room systems utilize "clean agent" technology, which means the suppressing agent is a gas which deprives the fire of oxygen, but leaves no residue and does not harm or contaminate equipment. Because of the walk-in nature of our engine room, we chose an agent that is also occupancy-safe; this is technology with which I am intimately familiar because it is exactly the same technology we used in the computer and telecommunications industries; I've managed the installation of dozens of these systems during my career.

I could spend several paragraphs here giving an overview of these systems for marine use, but instead I will refer you to this great overview article by good friend, fellow gear-head, and marine technical consultant par excellence, Steve D'Antonio. As Steve so clearly describes in that article, sizing the system correctly is critical, especially when using occupancy-safe gasses, which become toxic above a specified concentration (and, conversely, become ineffective below another specified concentration).

Sizing the system involves accurately measuring the volume of the space, in cubic feet. Vector's enormous engine room (we have perhaps twice the ER volume of other similarly sized boats) is irregularly shaped, even above the relatively flat sole (floor, in boat-speak). I had to divide this space into two sections; the taller, wider section forward of the saddle fuel tanks, and the lower, narrower section aft.

The bilge, or space below the sole, is even weirder, and I had to divide that into four sections, modeling each as an irregular prism and using solid geometry to calculate the volumes. Adding the resulting six numbers together provided the total volume of the engine room space, which turns out to be 1,108 cubic feet. Extinguishers are sized to the nearest 50 cubic feet, and you generally order the nearest size that is the same as or larger than the space to be protected, or, in our case, an 1,150 cubic foot system.

There are really only two vendors of pre-engineered (a fancy word meaning "not custom made"), USCG-approved systems in the US, Sea-Fire and Fireboy. After sizing the system, I consulted the web sites of each vendor to find their system size charts. There, I learned that the Fireboy system was packaged as a 10" diameter bottle, 21.4" tall, and that the Sea-Fire system came in a 10" x 26" bottle. That extra 4.5" of height turns out to be a really big deal in our engine room, as will become clear later, so I decided to go with the Fireboy system, even though I prefer the Sea-Fire units for technical reasons (better nozzle technology, and a shipping pin that the Fireboy lacks).

After shopping around, I was able to find a Fireboy unit for just under $2,500, plus another $70 for the manual release cable. That's a good deal; the clean agent itself, 47 pounds of HFC-227ea (sometimes known by the Dupont trademark FM-200) represents the major share of the cost, with the bottle, nozzle, and pressure gauge being relatively inexpensive off-the-shelf items.

An unplanned six-week stay in Chattanooga was just what we needed to get this done, as there is about a two-week lead time on delivery and I figured on a week or two to complete the installation. I placed the order and followed up with the retailer as the system was assembled and filled by Fireboy. As I mentioned here in a previous post, we were thus very disappointed when the system arrived in a 10" x 27.5" bottle -- even taller than the Sea-Fire system.

Oops... too tall.

It turned out that Fireboy had made an engineering change but failed to update the size chart on their web site (or the downloadable installation manuals, which were three full revisions out of date). Fortunately, the 1,100 cubic foot system was still packaged in the smaller bottle, and Fireboy's engineers agreed we could use the smaller unit for our 1,108 cubic foot space (1,125' is the cutoff). It was a simple, if tedious, matter to make the exchange. It cost me another week, but we did get $70 back for the smaller unit.

With the new bottle and the correct installation instructions in hand, I was finally able to start the installation. Step one was to finalize the installation location, from a choice of three spots I had picked out earlier, albeit with outdated literature on hand.

You really want to get these systems as high as possible in the space; heat rises, and the automatic discharge depends on a little actuator in the nozzle melting, at 175°F. Ideally the nozzle should be just below the highest point in the ceiling; installation directions call for it to be no lower than 20" below that. Because our engine room is taller in the front than the back, ideally the bottle would mount in this area, preferably to the forward bulkhead.

Looking down at the business end. Clockwise from top: discharge nozzle with thermal actuator and pinned manual release lever; pressure switch, charge/fill port, pressure gauge.

Sadly, most of the real estate on this bulkhead is occupied by the fuel polishing system and the stabilizer heat exchanger, both of which were added by the same cut-rate contractor who had little regard for the possibility of ever wanting to install anything else later. I found one nearly ideal spot which would fit the bottle, but it turned out that the lower bracket is considerably larger than the bottle itself, and I could not make this location work without moving one of the other systems, a very big job.

The other location, also in this taller section, was just aft of the door on the port bulkhead. This was also a great spot, with easy access behind the bulkhead to fasten the bolts (the forward bulkhead would have required surgery on the nicely finished aft wall of the master stateroom), but, here again, that pesky lower bracket would require me to cut a 4"x1.5" notch in a steel structural support for the bulkhead. Messy, tedious, and possibly making a weak point in that bulkhead.

Ultimately I settled on the third choice, against the aft bulkhead, between the day tank and the generator enclosure. While not as high as I would have liked, the nozzle is just below the level of the vast majority of the engine room ceiling. It's also much closer to the most likely fire-producing items in the room -- the generator, and the exhaust/turbocharger for the main engine. The space here is otherwise unused and out of the way, and there is easy access via the tiller flat (lazarrette) to fasten the bolts.

Insulation removed. You can see where I've had to grind down the weld seam. The white paint is leftover from the Racor filter installation below; I had to loosen the clamp and move the filter housing down about an inch to provide adequate service clearance for it.

The next step was to remove the thick fiberglass insulation from the installation location. This stuff is held to the walls (and ceiling) by being impaled on a series of spikes that look like nails spot-welded to the wall. I only exposed one spike in this process, which I bent out of the way so I could put some of the insulation back when I was finished.

The bulkhead here is 1/4" steel plate, and when I got the insulation off I found a heavy weld seam where two plates were joined. This seam, plus a slight misalignment of the plates, kept the brackets from lying flat against the wall, and there was no way to reposition to avoid the seam. Out came the 10,000-rpm mini-grinder, and an hour or so of careful work later, the seam was flat, but grinding dust was everywhere.

The ten 1/2" mounting holes as seen from the back side of the bulkhead, in the tiller flat.

With the brackets now able to lie mostly flat (there was still a small gap at one end due to the misalignment, which I handled with some washers), the next step was to drill the mounting holes. There was no specification for fasteners other than that they should be of "appropriate" size. Since the hardware came with 0.56" through-holes, I opted to use 1/2" stainless bolts; with some 75 pounds hanging from a moving wall I wanted plenty of reserve holding power.

I seldom use a corded drill any more; I can't remember the last time I used mine (well before the boat) and I sometimes wonder why I keep it. But drilling ten 1/2" holes in quarter plate is not a job for a cordless. I completely used up my 1/2" cobalt drill bit in the process. No template was supplied, so I had to measure locations as best I could to get all the brackets to line up.

Mounting area primed and painted.

Like many parts of the engine room, this wall was never painted, and was still mostly in red oxide primer. I wanted a more finished look, but also I needed now to protect the steel I just exposed with grinding and drilling. I used some Rustoleum primer I had on hand and the remains of a can of white Krylon to cover all the exposed area.

Getting the brackets bolted to the wall was a two-person affair, with Louise turning the ratchet from the engine room side, while I wrenched on the bolts from the tiller flat. My alignment was pretty good; when I set the bottle in the bottom "cup" it was off of plumb by just a fraction; tightening the straps makes it right, but the bottom of the bottle now touches the cup in only one spot. I could probably get it flat by loosening all the bolts and "adjusting" the cup portion with a mallet, but I think it's unnecessary with the straps properly tightened.

Weighing the bottle. At 72.3 lbs, just 1% over rated weight. I set the tare on the scale with the rope in place.

Just before hefting the bottle into place, I hung it from my digital crane scale to weigh it. This is a baseline; every year I will need to disconnect the bottle, take it down, and weigh it like this. If the weight drops more than 5% from the gross weight on the label, the unit requires service. At this moment it actually scales out above label weight; I think the factory tends to overfill by a small amount.

Rating label showing proper agent fill and tare weight.

Heaving 73 pounds up above chest height while kneeling in a cramped space was probably the most difficult part of the entire process. Since I have to repeat this step annually, I may build myself a little platform to rest the bottle on while I tip it into the bracket.

You can see a bit of red gaffer's tape peeking out around the steel straps. The instructions suggest isolating the bottle galvanically from any metal structure; this tape, along with some between the strut and the bottle and some on the bottom rim, achieves that isolation. I honestly don't think it's a big deal; the steel of the bottle and the steel of the boat are nearly identical galvanically.

Bottle mounted and secured. Racor filter has been lowered. I later reinstalled some of the insulation on either side.

With the bottle thus mounted, the engine room is now protected; if the temperature at the nozzle reaches 175°F the bottle will discharge, and, in fact, there's no way to disable this. It's also now safe to move the boat, with all the heavy bits properly secured (we didn't want to get under way with a fully armed fire bottle loose on deck once we unpacked it). Mercifully, I hear you thinking, because this post is already too long, but, alas, no, we are only half done.

There are two other parts to the project, one mechanical and one electrical. The first is straightforward: our bottle is equipped with an optional manual release. Certain fires can do plenty of damage and even produce plenty of toxic smoke long before temperatures at the ceiling reach the magic 175°F. If you watch the video I linked at the beginning of this article, you will see this fire was extinguished manually long before the automatic system installed on that vessel could discharge.

We have smoke detectors in all compartments, including the engine room. They're all linked together so that if any one goes off, we'll hear it on the unit right here in the pilothouse. The smoke detector will give us much earlier warning than the heat-activated discharge, and the aforementioned video camera will let us see right from the pilothouse if things have already progressed too far to enter the engine room to fight it with a handheld. The manual release will allow us to discharge the bottle without going below.

Manual release in salon. Remove pin, then pull handle.

Manual release cables are available up to 100' long, and we could easily have ordered a 50' model to mount the release handle at the helm. But that's very expensive, and longer cables with more and tighter turns pose a greater risk of binding, potentially at an inopportune moment. I ordered a 14' release cable, enough to get to a convenient place in the galley or salon from any of the three potential mounting sites.

Inside the cabinet, above the DirecTV receiver. Bends in the cable need to be large-radius as shown.

With the bottle at the very aft end of the engine room, the best location for the manual release handle is at the port side cabinet at the aft end of the salon, what we call the "entertainment center" because there is located all the A/V electronics except for the TV itself and its sound bar. This cabinet afforded the 18" or so of depth required for the release mechanism, and put the handle in a very visible and easily accessible location, yet out of the way of any potential unintended contact. It's just eight steps for either of us from where we sit in the pilothouse, and even closer to our main seating when the generator is in use.

Hole through deck, as seen from below, in engine room. Water lines go to deck shower.

I did have to drill another hole through 1/4" steel plate for this cable, right between two existing penetrations for the water lines to the aft deck shower. Once that was done and deburred, it was a simple matter to run the cable, install and test the pull handle, and connect it to the bottle. Fiddling with the manual release while standing right in front of the discharge nozzle is a high pucker-factor experience, however.

Cable coming through floor into cabinet.

The electrical side of things is somewhat more complicated. Clean agent suppression relies on the gas concentration in the space being above a specific concentration, in this case 6%, and remaining there long enough to ensure the fire is out and does not reignite. That means the protected space needs to be fairly effectively sealed off, but it also means that there can be no equipment removing gas from the room and/or bringing fresh air in to replace it.

In order to accomplish this, the fire bottle is fitted with a pressure switch, the purpose of which is to shut down every air mover in the engine room. That includes the engines, which draw their combustion air from inside the room and exhaust it out of the boat through the hull.

We have one exhaust fan in the engine room. This is an AC-powered Dayton model (Grainger's house brand) installed by a contractor during the last owner's tenure. When we got the boat, it was connected via a simple on/off switch to a circuit powered only when the generator was running or shore power was connected. Early on, I rewired it to an inverter circuit, so we could run it under way without needing the generator running.

Way back in March, I built a fan control system so that the fan would run whenever either the generator or main engine was running, without one of us having to go down and turn it on. A three-position switch allows us to choose between this automatic mode or to force the fan on or off regardless of engine status. When I designed this, I did it with the future fire bottle installation in mind, such that it was a simple one-wire change to allow the fire system to shut the fan down.

The engines were another matter. Gasoline engines do not need any sort of modification, because the suppressing agent stops them cold just as soon as they get a big gulp of it. But diesels will keep running even when breathing clean agent gas, and so need to be shut down electrically when the bottle discharges.

The pressure switch on the fire bottle is a simple SPST, NO type switch, which closes upon a set pressure, I would guess in the neighborhood of 350 PSI. So the switch is closed at all times so long as the bottle is not discharged, and opens fairly quickly once the gas is released. It is intended to be in the ground end of the operating circuit for anything needing to be shut down, and it is also the ground for an indicator lamp at the helm, which is included in the kit.

Indicator installed at the helm (and literally behind the helm wheel). Green light means full bottle; dark means discharged.

In order for this "ground-to-operate" type of system to work, it's necessary to find a circuit, or else add one, on the engine's control system, where interrupting the ground will cause a shutdown. On the generator that seems fairly straightforward: there is a "run" signal that keeps the normally-closed fuel solenoid open; breaking that run circuit will close the fuel valve and stop the engine. All well and good, except the fuel solenoid itself is grounded directly to the engine -- it threads into a hole in the injection pump.

Noodling over the generator schematics (and let me just say that Northern Lights generator run/stop controls are unnecessarily complex), I found a relay (one of four) in the control box that supplies the power to the fuel solenoid. Clipping the ground from this relay and re-routing it to the fire system pressure switch was all I needed to do to make the generator stop when the bottle discharges.

The main engine is a much more challenging problem. By design, once the engine is running, you could literally rip the entire wiring harness from it and throw it, as well as the batteries, overboard, yet the engine will continue to run. The engine is stopped by supplying 12vdc to a "stop solenoid," a fuel valve which is open by default and closes only while power is applied.

New main engine stop relay, below fan control box.

In order to make the mere interruption of a ground signal cause the engine to stop, I had to add a relay that would supply power to the stop solenoid, but only when the engine was already running and the signal was interrupted. To do this, I make the assumption that the engine key switch is "on" when the engine is running, even though the engine continues to run if the key is turned off. I use this signal to power the stop relay if and only if the fire bottle switch opens. It's not a bad assumption, inasmuch as all the engine instruments and alarms (tach, oil pressure, water temperature, voltmeter, and hourmeter) stop working if the key is turned to the "off" position.

A set of diodes keeps any one of these three systems from falsely grounding the others. The indicator lamp, also connected here, is an LED, so no additional diode is needed. When the pressure switch opens, the fan, generator, and main engine will all stop, and the indicator lamp will go out. It took several hours to design this fail-safe system, and a few hours to run all the wires and wire it all up.

Normal/Test/Override switch in engine room.

Before these diode-steered ground signals get to the actual pressure switch, they pass through a three-position switch, with safety cover, in the engine room. The normal position passes the signals on to the pressure switch. A "test" position interrupts the signals, providing a functional test of the shutdown system. And an "override" position bypasses the pressure switch and sends them directly to ground. This allows the systems to be restarted if and when the bottle discharges, so that the gas can be exhausted from the room, and engines, if still operational, can be used.

With safety cover open. Closing the cover forces the switch into the normal "active" position.

The final piece of the installation consists of an emergency override switch at the helm, also equipped with a safety cover. This switch interrupts the "run" signal from the key switch to the engine itself. The engine will continue to run (or could be restarted if already stopped) but the alternator will be turned off when this switch is opened. This is critical in the event that close-quarter maneuvering is required even while a fire is happening.

Fire override switch at the helm allows main engine to keep running if needed.

All told it took me perhaps 30 hours to install the system. In addition to the $2,600 for the bottle, cable, and shipping, I spent less than another $50 on miscellaneous wire, switches, covers, and mounting hardware. A turnkey professional installation would have come in somewhere north of $10,000.

As soon as I was finished I called our insurance agent to let him know we now had a system installed, and to see if there was any policy discount to be applied. The underwriter, who was very keen to know if we had such a system back when they quoted the policy, declined to provide any further discount. But the peace of mind is priceless; short of the ship sinking, there is no greater danger aboard than a fire at sea.


  1. Good article. I am astounded that your insurance underwriter wouldn't discount your premium after installing such a valuable item. Prahaps due to your 'DIY' installation? I would be shopping for a new insurance company.

    I do enjoy reading good tech articles like this one. Keep 'me coming!

    1. Thanks for your comment. We did not mention at all that it was a DIY project, so that was not a factor. We just said that a system "had been installed."

      Shopping for new insurance is not to be taken lightly with a boat such as ours. This particular issue is not enough to tip the scales.

  2. Nice job Sean, keeps the engineering mind active.

  3. First, I love the more technical posts.

    Two questions: (1) Did you mean 0.056 through hole or 0.56?
    (2) How much of a concern are explosive atmospheres? I don't know much about this kind of marine system, but if you have a comustable gas in your engine room, your diesel engine will keep running even after the fuel pump is deenergized. You would want an air I take shutoff valve.

    1. (1) Good catch -- 0.56" is what I meant to type. Fixed now.

      (2) Not a concern at all. Nothing in the ER gives off combustible gas. Gasoline-powered boats have far more issues (and thus safety standards) than diesel power such as we have.


Share your comments on this post! We currently allow anyone to comment without registering. If you choose to use the "anonymous" option, please add your name or nickname to the bottom of your comment, within the main comment box. Getting feedback signed simply "anonymous" is kind of like having strangers shout things at us on the street: a bit disconcerting. Thanks!