Domestic Plumbing for Sailboats



    The first time I tried to take a shower onboard I thought I would die. The water came out in bursts as the electric water pump cycled on and off and it was impossible to adjust the water mix for a stable temperature. One moment the water was freezing and the next it was scalding... I ended up washing myself out of a bucket I filled with water at a comfortable temperature.
    In the galley sink the problem was very similar. The pump cycling was very annoying and it was impossible to get a stable temperature. For the pump to not cycle you needed to open the faucet enough to get a large flow of water... that splashed all over the place. Clearly the plumbing system required some improving.
    I asked some fellow boat owners but they were not much help. It seemed to me they did not think this was such a big problem. I came to the conclusion they had learnt to live with the problem. I also asked at boating and RV stores but they were not much help either. What most impressed me was how willing people are to give advice even though they know little or nothing. But in their willingness to help they will give whatever advice they can think of. The most common advice was to install an accumulator. I heard this so many times that I now feel I should explain what an accumulator does and what it does not do. But first...


A word about pumps

    There are two basic types of pumps used in a sailboat: reciprocating (alternating) and centrifugal (rotary). A manual bilge pump is an example of a reciprocating pump. It has a chamber with a volume that increases with one stroke and decreases with the reverse stroke. Two check valves let water in from one side and out to the other side. The pumping chamber can change volume by means of a piston or by means of a diaphragm.
    The manual pump fitted in the head is of the piston type and if you have ever assembled one (each valve has it's own name: Flapper, Flipper, Joker, Stroker, Jigger, Jogger...) you know what I am talking about.
    If you have used a manual bilge pump of the diaphragm type you know and understand why the water comes out in spurts and not in a continuous flow. An electrically powered diaphragm pump is essentially the same thing. The rotary motion of the motor is converted to alternative motion to move the diaphragm and with every oscillation the pump pushes a small amount of water. This "pulsing" can be very much diminished by having one motor operate three or more chambers out of phase so that when one camber is actually pumping, the others are intaking. Three chambers shifted 120 are enough to cancel most of the pulsing.
    These alternating pumps are also called positive displacement pumps because with every stroke it takes a volume of fluid and *positively* pumps it out the other side. What this means is that for every cycle it will pump pretty much the same amount of fluid. Still, because the diaphragms are not pistons and do have some flexibility and because the motor will slow down with load the flow will decrease with increasing pressure on the output side. Because of the "positive displacement" aspect these pumps are generally self priming. They will pump the air through until the water arrives.
    Nevertheless, if the output flow is restricted the pressure would build up to the point where the pump would be damaged so, for domestic water supply, a pressure switch is fitted on the pump that automatically switches the motor off when a certain pressure is reached and switches it on again when the pressure falls below a certain threshold. This is what makes the pump cycle on and off.
    A centrifugal pump requires the very fast rotation of the impeller and for this reason manual pumps are not of this type. In the small sizes used in boats, rotary centrifugal pumps do not provide much pressure but will provide more flow and are more forgiving of dirt in the water. For this reason electric bilge pumps are usually of the centrifugal type and circulator pumps that do not need to overcome much pressure are also of this type. Note that centrifugal pumps are not self priming because they rely on the mass of the fluid and the mass of air is a very small fraction of the mass of water.
    It should also be noted that an alternating pump has check valves built in while a centrifugal pump does not. It may be necessary or advisable to fit a check valve to a centrifugal pump. For instance, I have seen the following problem: The float activates the bilge pump which starts to pump water up the exhaust hose. Once the level has fallen the pump stops but now all the water left in the hose drains back into the bilge, maybe starting the pump again. A check valve will prevent this problem. Another problem is that if the outlet should go below water (by heeling or the boat sitting lower in the water) water can find its way into the boat. So it is a good idea to fit a check valve to a centrifugal bilge pump. On the other hand I have seen a check valve which had stuck closed after some time of no flow and the pump pressure could not force it open so the pump burned out and the batteries went dead. This is why you continually need to inspect things are in proper working order.
    There are other types of pumps on a boat. A macerator pump is a combination of a grinder and a pump. The water pump on the auxiliary engine is of a special type with a flexible impeller. These would deserve chapters of their own but our focus here is on domestic water systems and so let us return to electrically operated diaphragm pumps.
    These are the type commonly used for domestic water on a boat because for small sizes they provide more pressure but this pump does not push the water continually through but rather in short, quick, pulses which repeat many times per second. This pulsing (vibration) can be annoying and even destructive and this is where the accumulator comes in.


Accumulating... knowledge

    We have seen that an alternating pump gives a pulsating output which is annoying and can even be damaging. An accumulator is simply a chamber that has some air trapped which acts as a cushion that absorbs the pulsations of an alternating pump. An accumulator (which should be installed as near to the outlet of the pump as possible) can do much to alleviate this pulsating effect. This is what an accumulator can do.
    If a diaphragm pump with an only chamber is giving a steady output of 2 GPM (125 cc/sec) it really means it is pumping about 2.5 times that (5 GPM) during 40% of the time and nothing during the other 60% of the time. If the cycle is repeated 20 times per second it means about 1/5 ounce of water is pumped in each cycle. The accumulator takes this water pumped in one stroke and spreads it over until the next stroke. The amount of water accumulated is very small, just a fraction of an ounce.
    Although they sell accumulators for this purpose it is worth noting that any chamber or length of pipe of sufficient volume and filled with air will do the same job. A commercially purchased one will have a rubber bladder to prevent the air from escaping but a length of tube in such a position that the trapped air will not escape will do the same job.


A better diaphragm pump

    There are pumps of the alternative type but, rather than one large chamber, they have several smaller chambers which work with offset phases and the pulsating effect is minimal because the phases overlap. These pumps are much better suited for pumping domestic water because they do not produce as much pulsation and they also tend to be much quieter than the type with a single diaphragm. But because they have much smaller valves they are only suited for water that is clean and a strainer or filter should always be installed in the input line.
    A pump of the type with a single pumping chamber is more inefficient and noisy but generally will have larger check valves which will allow some dirt to get through without problem. This makes them better suited for pumping iceboxes, shower sumps, bilges, etc. Still, for purposes such as pumping bilge water, a centrifugal pump is even better because it is more forgiving of dirt and pumping the bilges does not require the high pressure provided by a positive displacement pump.
    But what an accumulator cannot do is make any difference in the slow changes in pressure that happen over several seconds as the pump cycles on and off. The pressure in the system will slowly rise until the pressure switch cuts off and then slowly fall until the switch starts the pump again. So much for the advice I was getting.
    If we are using a flow of water which is below the steady flow of the pump, then the pressure in the system will rise until the pressure switch cuts the pump off. Then the pressure starts dropping until the switch cuts in again. In the case of my pump the switch disconnects the pump when the pressure reaches 44 PSI and starts the pump again when the pressure has dropped to 22 PSI. A larger accumulator will make the cycles longer but it cannot change the fact that the pressure will oscillate between the upper and lower limits. Which I find very annoying.
    The paths of the hot and cold water have different resistance (as they are bound to) what happens is that the mix of hot and cold water will vary with the pressure. Of course the flow will vary as well. So what we have is a situation where both the flow and the temperature are oscillating. This does not make for a comfortable shower or, indeed, even for using the sink.
    Given that I could not find much help, it seemed I would need to study and resolve the problem myself. It took me quite a few hours of studying and head scratching but once I had a clear understanding of the technical issues involved, the solution turned out to be quite simple
    Rather than just give you the solution directly, I would rather explain the process so you may also understand it and this way you can better adapt it to the specific installation on your own boat or RV.


Practical Experiments

    In my case, the 12 volt electric water pump is a Shurflo model 2088-423-344. It has three pumping chambers and it is rated nominally at 2.8 gallons per minute (GPM) and 40 PSI. Note that this does not mean it will pump 2.8 GPM at 40 PSI. What it means is that it will pump 2.8 GPM with an open discharge (0 PSI) and that the electric shutoff switch is set somewhere in the neighborhood of 40 PSI. This type is the most common on boats and well suited to the job.
    I hooked up a pressure gauge and with a gallon jug and chronometer proceeded to make some measurements. With a totally open output (= 0 PSI) it pumps about 2.8 GPM. At 40 PSI the flow is about 1.5 GPM and for simplification purposes we will assume the load graph is a straight line since it comes pretty close in the range we are concerned with.
    In graph 1 we can see that for flows over 1.3 GPM the pump would work continuously without cycling. So, ideally, one would want to increase flow to stop cycling. On the other hand, one needs to save water on a boat and even if that were not the case, as I say, trying to rinse a dish under a faucet that is shooting out 2 GPM is a way to take an unintended shower because water splashes everywhere. In practice most water usage is done at rates between 0.5 and 1 GPM and this is the range where we want a comfortable and stable output.
    For flows under 1.3 GPM the pump cycles on until the pressure reaches about 45 PSI and then switches off until it falls to about 22 PSI when it switches on again. So with or without an accumulator, there is no way around the fact that the pressure will oscillate between 22 and 45 PSI if the flow is below 1.3 GPM.
    Even for a shower, where I need more water, with a water-saving showerhead I can shower comfortably with 1 GPM and even less. And in the galley sink I would rarely want a greater flow. So my first experiment was this: While I used the galley sink I would leave the bathroom sink slightly open. This "spilled" water increased the total flow through the pump so it would not cycle while the flow I was actually using myself was smaller. This works but, of course, is a waste of water.


Practical solution #1

    But what if this spilled water, instead of going to waste, went back to the water tank? It would not be wasted and the pump cycling would be avoided. So my first idea for a solution to the cycling problem was this: I would install a pressure relief valve that would open at (say) 30 PSI and allow a small flow back to the tank. It would have to open at above the lower cutoff point (22 PSI) so the pump would stay switched off and the flow back would have to be restricted so the pressure would build up enough to switch the pump off.
    So now the flow-pressure graph would look like graph 2. Below 30 PSI (which is to say for flows above 1.8 GPM) the graph looks exactly like before because the pressure relief valve is shut.
    But once the flow is restricted below 1.8 GPM the pressure relief valve opens and allows some water to escape back to the tank or just back to the low-pressure entrance of the pump. This flow has to be restricted though so that with diminishing flow the pressure will build up to the cutoff point. With too much relief this would never happen and the pump would work without ever stopping. In graph #2 we see that now we can use as little as 0.3 GPM and the pump will not cycle.
    This is an elegant solution to the pump cycling problem and it certainly makes the use of water much more comfortable. The only (very minor) consideration is that the pump will work continuously when it would otherwise be cycling and this represents a small increase in electric consumption.
    I do not understand why the pump manufacturers do not build this into their pumps because the added cost is negligible and the convenience it adds is well worth it.


Practical solution #2

    Now that I had a solution I had created myself another problem: where to find the parts I needed. I intended to use a small needle valve to adjust the relief flow but the pressure relief valve proved very difficult to find. While I looked for it I kept studying the problem.
    A small relief orifice that connected directly the output side to the input side would make the graph look like graph #3. While water is being consumed this is very well but the problem arises when no water is being consumed because then the pump is continually cycling to build up the pressure that is being lost. This was clearly not an acceptable solution. I toyed with some ideas that would involve some electronic detectors and controls to but this solution was seeming more complicated than finding a pressure relief valve. So... I kept thinking and scratching my head...
    Now follow me for a moment while we look at the pump itself. A diaphragm pump is basically a chamber of varying volume with two check valves. When the volume of the chamber increases, the input check valve allow water to enter the chamber and when the volume of the chamber diminishes the outlet check valve allows the water out to the high pressure side.
    In the diagram the water flow is marked in green, the oscillating diaphragm in blue and the check valves in red. This is only a schematic diagram, simplified for our purposes, and not a true representation of how the pump is built. One interesting point is that the input check valve is part of the diaphragm itself and not mounted in front of the input port.
    So it occurred to me that I could have a relief orifice bypassing the input check valve (this is, connecting the input side to the main chamber) and this would achieve the effect I was looking for without causing that problem. The pressure from the high side would not be lost through this orifice which only has any effect when the pump is at work.
    I studied the pump and saw that, looking into the input side I could see a wall that separated the low pressure input side from the intermediate chamber. I could drill a small hole through that wall without even disassembling the pump (purple arrow in diagram). I decided to try it figuring the worst that could happen would be that I would have to plug it with epoxy.
    After endless calculations as to flow, pressure etc, I drilled a very small hole and it worked but I wanted more effect so I gradually drilled larger until at last I settled for an orifice drilled with a 2.2 mm bit (0.085"). With this the system works beautifully. I can use as little as 0.3 GPM without the pump cycling.
    But, of course, there is always some unexpected problem. I was so happy with the way this worked and all it had cost me was drilling a tiny hole!! Well, it did work fine for many weeks but one day the tank emptied and the pump sucked some air. It never occurred to me that now the pump was no longer self priming. The tiny orifice presents quite a resistance to the flow of water but almost none to air. So now I have a new problem: when the water tank empties and the pump sucks air, it will stop working. It can handle small amounts of air bit not a lot. Once the pump has sucked a lot of air in, it has to be primed again. I have found I can do this easily just by sucking once from the faucet and it is so easy I have not even thought of providing any other solution but this is because the pump is right by the water tank. If there were a longer run it would take a little more to prime it.
    I do not think anybody would consider this a serious problem but you may want to take it into account. The ideal installation, of course, would have the pump below the tank level so it would not need priming. Another solution would be to fit a small, hand-operated, priming pump.
    Before you drill any holes in the body of the pump you can get a fair idea of what the effect will be just by letting another faucet a bit open so it will let some water run, say 1/2 GPM (remember? this is how we started). Now use another faucet normally and see how the pump does not cycle.
    An observation I would make and which applies to both solutions: For domestic water consumption one generally does not need high pressures and high flow. I have used solution #2 on my boat Callisto for two seasons now and it works beautifully. But if you use the pump for washing down decks or other applications where you want maximum pressure and maximum flow at the same time, then, as you can easily see, these solutions are not adequate. In such case I would recommend installing separate pumps.

And another observation: Always install a water filter before the pump. I am always amazed at the amount of stuff that my filter catches. I have no idea where it comes from but it's there.


Electric addition

    By the way I will mention that I also experimented with electronic controls for the pump's electric motor. It just seemed the pump was more powerful than I really needed. After much experimenting with both switching and linear devices I came to several conclusions.

One is that you can lower the power of the pump by a certain percentage by using a switching or analog device but you cannot lower it too much or the pump will not reach the shutoff pressure. It will stall and possibly burn the motor.
    Another is that all my digital switching stuff really wasn't worth it. A resistor in series with the pump is all that is needed. In my case and with my pump, a series resistor of 0.26 ohm provides a very nice "low power" position for the pump. So now I normally have this resistor in series and only short it with a switch when I want greater flow or pressure. You would have to experiment with your particular pump but as a starting point I would suggest a value in ohms equal to the number 2 divided by the rated amps of the pump. So, for a pump rated at 4 amps a resistor of 0.5 ohm is a good starting point. Take into account it has to dissipate a good amount of heat (in this case 8 watts). If the pump slows too much and cannot reach the shutoff pressure, then diminish the value of the resistor. On the other hand, if it does not have enough effect you can increase the value.
    If you want to get fancy you can design some system that would automatically switch between high and low power depending on your requirements. For example it could only switch to "high" when the shower is being used. This would depend very much on personal preference and use of the water system.


Water heaters

    At this point we have solved the pump cycling problem. Having a constant pressure and flow make it so much more easier to adjust the mix of hot and cold water to obtain a comfortable temperature but it is still tricky. In theory, once we have adjusted the mix, if nothing changes, with the pressure remaining constant, the temperature should not change.

In practice though, achieving the right temperature is tricky. Let us look at some rough numbers. If the cold and hot water temperatures are about 77F and 195F respectively, they need to be mixed roughly 3 parts cold to 1 part hot to obtain a comfortable shower but any substantial deviation means a scalding temperature.
    My water heater holds 6 gallons which is pretty standard. This means that under those conditions it would provide a total of 24 gallons of warm water. It would be better to have a water heater that held 24 gallons directly at a usable (lower) temperature but, in general this is not feasible on a boat.
    When it comes to water heaters, everything else being equal, bigger is better. It not only means that we have more hot water but it also means heat is transferred more efficiently from the motor heat exchanger and that the temperature will remain more constant while we use hot water. The only reason I can think of (besides cost) to install a smaller water heater is the placement of the weight. Note that I say "placement" and not "addition" because the weight of the water will be carried anyway but if carried in the hot water tank it is generally in worse place for the balance of the boat than if carried in the cold water tank.
    After running my auxiliary motor for less than an hour, the temperature of the water in the heater has gone as high as it will go and will not absorb any more heat. In cold weather I find it a waste to be dumping heat into the sea and if I am running the motor for long periods I will empty the hot water in the heater into the cold water tank and so fill the heater with cold water again. This means the water in the cold water tank will be warmer and so less water from the heater will be needed in the mix to obtain the desired temperature. While I do this by putting a hose from the faucet to the tank inlet, it would be easy to provide a valve for this purpose.
    It should be understood that the heat exchanger in the heater circulates hot coolant from the motor and it is practically impossible that the water in the heater would reach the boiling point. On the other hand, the electric element in the heater will heat the water until it boils and evaporates and so a thermostatic switch is provided that will cut off the heat when the temperature reaches a certain point.
    I have seen boat water heaters that have a heat exchanger but no electric element. If the heater is otherwise working in a satisfactory manner, rather than replace it with one that has an electric heating element, I would suggest installing the heating element (and thermostatic switch) in a separate small container with a small circulation pump that circulates the water through the heater. This can easily be accomplished by installing the electric element, the circulator pump and a check valve in parallel with the tank of the water heater.


Thermostatic mixing valves

    Water that gets too cold when one is taking a shower is a very uncomfortable nuisance but if it gets too hot it is also dangerous. At home I have resolved this by lowering the temperature of the thermostat in the water heater but there I have a water heater that holds 40 gallons preceded by another solar heated tank that holds another 60 gallons. I have set the thermostat to where the water feels comfortable with the addition of little or no cold water. But in the boat this is not a feasible solution because the water heater holds only 6 gallons which would rapidly be exhausted.
    The solution was to install a thermostatic mixing valve. It was not easy to find because I was looking in the wrong places (boating and RV supply stores) but I finally found it in a plumbing and heating supply company. These are sold and installed for several purposes and one of them is precisely to protect children or older people from scalding burns. I would recommend installing them at home too as a protection against scalding (they are required by the building codes for new construction in many jurisdictions which shows they are really necessary).
    The particular thermostatic valve I used is a Sparco Aquamix AM100L. It cost just under $100 which I found expensive but well worth the price. The output temperature can be adjusted within certain limits (90 - 120F). It automatically mixes the correct proportions to maintain the set temperature.
    I installed it right close to the water heater and it works beautifully. As the hot water in the heater is used up the temperature decreases and the thermostatic valve compensates for this too. As the temperature of the water circulating through the pipes is lower, heat losses are minimized.
    The only observation I would make is this: while no water is used in a long time, the water in the valve cools and so the valve opens the hot side. When you first open a faucet the valve takes a few seconds to react so you may get a few seconds of cold water that was in the pipe, followed by a burst of scalding water and then, once the valve reacts, the water temperature settles to its correct value. In my boat this is not very appreciable because this initial hot water is cooled by the pipe that goes all the way to the faucet. But if this is a problem in your installation the solution is simple: either stay clear of the water for the first few seconds or, if you are truly obsessed with saving water, open the faucet for a few seconds so the hot water reaches the mixing valve, close it for a few seconds while the valve reacts and then open it again for normal use.
Sparco Inc, 65 Access Rd, Warwick, RI, tel. 401-7384280,
Capitol Hidronic, 1311 S Fern St, Arlington, VA, tel 703-4168555,
Hughes Supply,



    Now I enjoy comfortable showers at stable and easily regulated flow and temperature. It has made all the difference for me.
    In my next article I will talk about other aspects of the plumbing system: why I chose polybutilene (PB) for the pipes, what kind of fittings I used and how I designed the system to incorporate simultaneously the onboard pressure pump and the dock pressurized supply.