Engines, Oil and Conspiracies

We have had engines in boats for more than a century. In a sailing yacht there is almost always exactly one. And still, on any busy summer weekend, if you leave the VHF on channel 16 you will hear it: engine failure, requesting assistance. Not a dismasting. Not a collision. The engine.

Let us admit something to each other, quietly, before anyone else is listening.

Most of us go to sea — real sea, sometimes ocean — behind a single point of failure. One engine. One gearbox. One raw-water pump. One shaft. (Unless you sail on two boats bolted together, which the rest of us call a catamaran, and which is a different conversation.) And we tell ourselves we are doing everything that can reasonably be done about it.

We are diligent. We change the oil and the oil filter every season, or every couple of hundred hours, whichever comes first. We change the fuel filters and drain the water separator. We carry a spare impeller, and most of us have even fitted one, once, upside down, in a seaway, while somebody held a torch. We check the anodes. We tension the alternator belt. We change the coolant. We clean the sea strainer. We look at the raw-water exhaust and count the beat of the water. And every single time, before starting, we pull the dipstick, wipe it on a rag we should have thrown away in 2019, put it back, pull it out again, and look at a line of black oil in bad light.

And we think: that is the most that can be done.

But maybe it is not. Maybe there is quite a lot more that can be done, and the strange thing — the thing that made me want to write this — is that nobody ever told us.

So is it a conspiracy? Is there a room somewhere in Gothenburg or Yokohama where engine manufacturers agree that the cruising sailor shall be told the pressure is fine until the moment it very much is not?

I do not know. I am not going to dig into it, and I would gently suggest you do not either. What I am going to do is explain, in some detail, what actually happens inside your engine, what your engine actually measures, why the gap between those two things is so much larger than anybody admits — and what the Galvanic Voice does about it, with real numbers off a real engine.

First, a Mechanical Lecture: Why a Diesel Lasts

A marine diesel is a deeply unglamorous machine, and that is its great virtue. Rudolf Diesel filed his patent in 1892 and had a working engine by 1897, and the thermodynamic idea has not changed since: compress air until it is hot enough to ignite fuel on contact, inject the fuel, and take the work out on the way down. No spark. No throttle plate. Air, compression, heat, fuel, work.

The consequence of that simplicity is that a well-treated marine diesel will run for eight, ten, fifteen thousand hours before it needs anything that could be called a rebuild. Commercial installations go further. This is not because the engine is delicate and lovingly nursed. It is because it is a robust thermodynamic machine built out of a small number of very well understood principles — and, above all, out of materials.

That last word is the one that matters, and it is worth slowing down on.

The reason a crankshaft made of hardened steel can spin inside a bearing shell made of a soft aluminium-tin or copper-lead alloy, thirty times a second, for fifteen thousand hours, without the two ever wearing each other away, is that under correct lubrication they never actually touch. The journal rides on a wedge of pressurised oil that it generates itself as it rotates — a hydrodynamic film a few thousandths of a millimetre thick, thinner than a human hair, carrying loads of tonnes per square centimetre. This is not a modern discovery. Beauchamp Tower stumbled onto it in railway-bearing experiments in 1883, and Osborne Reynolds wrote down the equation that governs it in 1886. Every plain bearing in your engine is an application of a paper published while Rudolf Diesel was still a student.

Which is the other half of the sentence, and the dangerous half. The moment the lubrication is insufficient, or — far more common, far more insidious — the lubrication is incorrect, the bargain breaks. And it does not break gently.

Oil becomes incorrect in a small number of well-known ways:

• Water in the oil. From a weeping head gasket, an oil-cooler core, a leaking exhaust elbow, or simply condensation in an engine that runs an hour a week. Water displaces the oil film, corrodes the bearing overlay, and destroys the additive package. In rolling-element bearing testing, contamination on the order of a tenth of a per cent of water has been shown to cut fatigue life dramatically (Cantley, ASLE Transactions, 1977) — a tenth of a per cent being a quantity you cannot see on a dipstick.
• Fuel in the oil. A leaking injector, a worn pump seal, or long periods of idling with incomplete combustion. Diesel is a solvent and a thinner: it lowers the viscosity, and viscosity is precisely the property that sets the film thickness. Most manufacturers condemn the oil at a few per cent of dilution. A few per cent is, again, invisible.
• Carbon and soot. Combustion products blowing past the rings, thickening the oil, loading the filter, polishing the cylinder bores.
• A filter going marginal. Every full-flow oil filter has a bypass valve, because a blocked filter must never be allowed to starve the engine. That bypass is a mercy and a trap: once it opens, the engine is running on unfiltered oil, and nothing on your panel says so.
• A raw-water impeller shedding its blades. Which is not an oil problem at all until, twenty minutes later, it is the only problem — because the coolant stopped being cooled, the oil went with it, and hot oil is thin oil.

Notice what all five have in common. Not one of them is visible on a dipstick, and not one of them will trip your oil-pressure alarm until it is far too late.

The Disaster Is Twice a Disaster

When a marine engine dies, it costs you twice.

The first cost is the repair, and on a modern cruising yacht that cost is dominated not by the parts but by access. Our own boat is a Hanse 588 — a wonderful boat, fast, beautifully laid out, and in which the engine is fitted with what the industry cheerfully calls “good access”. In practice the access is what I would describe as virtual. If that engine ever has to come out, I am fairly sure the first tool required is a saw, and the first casualty is the galley entrance. This is not a criticism of Hanse in particular; it is the universal geometry of a modern aft-cabin cruiser. The engine went in before the interior did.

The second cost is where it happens. An engine failure alongside is an expensive afternoon and a bad mood. An engine failure at sea is a different category of event, and the reason is that on a modern yacht the engine is not only propulsion:

• You are now sailing home — if you have the sails up, the crew, the sea room and the wind angle to do it. In a calm off a lee shore, in a foul tide, or in a traffic separation scheme, “we will sail” is a sentence with conditions attached.
• You have lost your battery charging. Which means, on a schedule that depends entirely on how well you provisioned electrically: autopilot, navigation lights, instruments, plotter, VHF, fridge, and the ability to call anybody about any of it.
• You have lost close-quarters manoeuvring exactly when you are most likely to need it — the harbour entrance at the end of a long day.

That is why “engine failure” is what you hear on 16. Not because engines are fragile. Because when the one you have stops, everything downstream of it stops too.

How the Oil Actually Gets Where It Is Going

To understand what your engine is not telling you, you need to know how the oil circuit is laid out. It is worth five minutes, because almost every failure mode above expresses itself in this circuit before it expresses itself anywhere else.

Oil lives in the sump. A pickup tube with a coarse strainer draws from the lowest point — which is why oil surging away from the pickup in a hard turn or a steep head sea produces a momentary pressure dip that means nothing at all, a detail that will matter later.

From there the oil reaches the pump, and the nature of that pump is the single most important fact in this article. It is a positive-displacement pump — a gear or gerotor pump, driven directly off the crankshaft. Positive displacement means it moves a fixed volume of oil per revolution, not per second. Turn the engine twice as fast and it delivers twice the flow. It does not “make pressure”; it makes flow.

The pump feeds the relief valve — a spring-loaded valve that dumps excess oil straight back to the sump once the gallery pressure reaches its setting. Then the oil cooler, a heat exchanger against engine coolant or against raw water. Then the full-flow filter, with its bypass valve. And only then the main gallery, the long drilled passage that runs the length of the block and from which everything is fed:

• the main bearings, straight off the gallery;
• the big-end bearings, through drillings inside the crankshaft itself — the oil goes into the crank at the main journal and comes out at the rod journal, carried there by the rotating shaft;
• the camshaft bearings, tappets and rocker gear, usually last in the queue and often through a restrictor;
• on many modern engines, piston-cooling jets that squirt oil at the underside of each piston crown;
• and on a turbocharged engine, the turbo bearing, which is a plain bearing spinning at a hundred thousand rpm and which is the first thing to die when oil supply falters.

Everything then drains back down to the sump by gravity, and does it again, forty or fifty times an hour.

Pressure Is a Delivery System, Not a Score

Before we go further, the single most important thing to understand about that number — and the reason a small change in it is so much more serious than it sounds.

The designer did not pick your engine’s oil pressure to look reassuring on a gauge. It was calibrated to reach every single friction point in the engine. Every one of them sits at the end of a drilled passage of a calculated diameter, at a calculated height above the sump, in a calculated position in the queue. The pump capacity, the bearing clearances, the drilling cross-sections and the relief-valve spring were all sized together, so that at the design gallery pressure each of those points receives the flow it was computed to need. Not approximately. By design.

And they are not equally easy to reach:

• The main bearings sit straight on the gallery. They are first in the queue and they are fine.
• The big ends are fed through drillings inside the spinning crankshaft, so the oil has to be pushed outwards against the centrifugal head of the rotating shaft before it arrives.
• The camshaft, tappets and rocker gear are at the top of the engine, furthest from the pump, usually last in the queue and often behind a deliberate restrictor.
• The piston-cooling jets, where fitted, are typically spring-loaded and only open above a threshold pressure.
• The turbo bearing is a plain bearing spinning at around a hundred thousand rpm, with a film to maintain and a great deal of heat to carry away.

So the gallery pressure is not a score out of ten. It is the supply head of a distribution network, and the network was engineered around it.

Which leads to the consequence that matters. When the pressure falls a little, it is not the case that everything in the engine becomes slightly less lubricated. The loss is not shared out evenly. It lands, almost entirely, on whichever point was designed with the least margin — the furthest away, the highest up, the smallest drilling, the fastest spinning, the one behind the restrictor. Spring-loaded piston jets simply stop opening. The rocker gear at the top of the engine, last in the queue, starves first. Everything else carries on looking perfectly healthy.

And at that one starved point, the bargain we described earlier breaks. The oil film thins past the thickness it needs, and the component slides out of the hydrodynamic regime — where the surfaces never touch — into the mixed and then the boundary regime, where the high spots of the two surfaces begin to make contact. Stribeck mapped this transition in 1902 and it has not moved since. It is a cliff, not a slope.

Two consequences follow, and both matter enormously:

One: pressure must rise with rpm. Flow is proportional to engine speed; leakage through a fixed clearance is not. So the same engine, healthy, makes low pressure at idle and high pressure at cruise. A single fixed threshold is therefore judging two completely different situations with one number.

Two: pressure must fall as the oil gets hotter. Leakage through a thin clearance is viscous flow, and viscosity collapses with temperature — an oil at 100 °C is several times thinner than the same oil at 60 °C. Hot oil escapes faster, so pressure drops. The same 2 bar is entirely healthy at hot idle and a genuine warning at cold cruise.

What “15W-40” Actually Tells You — and Why It Is Not a Slogan

Since viscosity is the property that sets the film, it is worth knowing that the number on your oil bottle is not marketing. It is a specification, it is enforced by measurement, and it has been an exact science for more than a century — the Society of Automotive Engineers published its first viscosity classification for crankcase oils in 1911, and its modern descendant, SAE J300, is the document every oil blender and every engine designer on earth works to.

A grade like 15W-40 is two independent guarantees, not one:

• The “15W” — W for Winter — is a cold guarantee. It caps how thick the oil may be when it is cold: a maximum viscosity in a cold-cranking test, so the starter can still turn the engine, and a maximum pumping viscosity at a lower temperature still, so that when it does fire, the pump can actually draw the oil up the pickup rather than carve a tunnel through it. A lower W number means a thinner cold oil and a colder safe start.
• The “40” is a hot guarantee, and it is the one that matters at sea. It says that at 100 °C the oil’s kinematic viscosity must fall inside a defined window — for an SAE 40, between 12.5 and 16.3 mm²/s. Not “about right”. A window, with a floor and a ceiling. On top of that sits a minimum high-temperature high-shear viscosity, measured at 150 °C under shear conditions chosen to mimic what actually happens inside a loaded bearing.

And the behaviour between those fixed points is not left to chance either. The way a mineral oil’s viscosity falls as it warms follows a form laid down by Walther and standardised as ASTM D341 in the 1930s — two measured points, and you can compute the viscosity at any temperature in between. It is the equation printed on every oil-company data sheet as a straight line on strange-looking paper.

Put plainly: the oil in your sump was designed to have a known, specified, measurable viscosity at every temperature your engine will ever see — and your engine’s bearing clearances, oil pump and relief valve were designed around exactly those numbers. The two halves were engineered to meet. This is not approximate. It is the most precisely specified fluid on your boat.

Which brings us to the point that I think is the most under-appreciated in this whole article.

The Galvanic Voice does not assume any of this. It measures it.

Look again at how the pressure arises. The pump delivers a flow proportional to engine speed. That flow escapes through fixed clearances, and the rate at which it escapes is set by the viscosity of the oil that is actually in the gap, at the temperature it is actually at. Rearrange that, and something rather elegant falls out:

That is what makes this real rather than theoretical. The Voice is not reading a label and trusting it. It is not looking up a grade in a table. It fits your engine’s own pressure–speed–temperature law from your engine’s own data, over thousands of hours, and then watches for that law to move. When the oil in the sump stops behaving like the oil the engine was designed around — because it has fuel in it, or water in it, or because it has simply had enough — the law moves, and the movement is the measurement.

The bottle tells you what the oil promised to do. The Voice tells you what it is actually doing, this afternoon, at 82 °C, in your engine.

And one more, which is the reason the whole problem is harder than it looks: when the relief valve is open, the pressure tells you almost nothing. On a cold engine, or at any decent rpm, the pump delivers far more than the bearings can leak, the relief valve opens, and the gallery pressure is pinned at the valve’s setting. You are no longer measuring your bearings. You are measuring a spring. Every cold-engine oil pressure reading you have ever admired was, in all probability, a reading of the spring.

What Your Engine Actually Measures

Now the short part. Here is the complete list of what a modern marine diesel reports about its own health, on essentially every engine sold in the last twenty years:

That is it. And notice: the engine is not measuring badly. Those four channels are exactly the right four. They are on the NMEA 2000 bus, in a standard message, twice a second, on the boat you own today. The sensors are good. The physics is a century old and completely settled. Every manufacturer has made slightly different choices about pump size, relief setting and clearances, but the machine is the same machine, and its designers knew precisely what pressure it should make at what speed and what temperature. They had to — they designed it that way on purpose.

The Paradox at the Heart of This

Read the oil-pressure row of that table again. A switch that closes at 0.3–0.5 bar — and note that it closes at that pressure whatever the engine is doing. One constant, for every rpm, every oil temperature, every load.

Now put our own engine’s real numbers beside it. This is the Volvo Penta D3-110 on our Hanse 588, at 2,098 hours, with the oil settled at its normal 80–83 °C — measured, not from a manual:

Look at the right-hand column, and then look at which way it runs. The harder you work the engine, the more useless the switch becomes. At idle it lets you lose three quarters of your oil pressure in silence. At cruise — where the bearing loads are highest, where the film matters most, where you actually spend your hours — it lets you lose seven eighths of it and says nothing at all. The correct pressure climbs with rpm. The switch does not move. The gap between them is the size of the blind spot, and it grows all afternoon.

The switch is not wrong, and it is not optional. It is a correct and necessary last line. But it is a certificate of death, arriving after the event it describes: by the time the gallery has fallen to a third of a bar, the hydrodynamic film has collapsed, the journals have touched, and the damage that determines the repair bill has already been done. It tells you nothing you can act on. It informs you that the accident has happened.

Meanwhile — and this is the part that should annoy you — the deviation that preceded it was sitting on the bus for weeks. Water in the oil, fuel dilution, a filter on its bypass, an impeller quietly losing blades: every one of them shifts the pressure-versus-speed-versus-temperature relationship slightly, and slightly is enough, because slightly is what “early” looks like. Caught there, these are cheap and easy jobs. A gasket. A cooler core. A filter. An impeller and an afternoon.

Caught at 0.4 bar, they are a crankshaft, a set of shells, possibly a block — and a boat with a hole cut in the galley entrance.

Why has nobody built the thing in between? Here I will disappoint the conspiracy theorists, because the honest answer is duller and worse. The thing in between exists, and has for fifty years — it is just not sold to us. Commercial shipping, power generation, mining and aviation all run continuous condition monitoring as a matter of routine, and industrial oil analysis programmes have been trending exactly these parameters since the 1970s. But that world samples oil into a bottle, posts it to a laboratory, and gets a spectrographic report back a week later, on a 200-hour or 500-hour cycle. It is a superb tool and it is discontinuous, retrospective and quite unsuited to a 58-foot boat whose owner would like to know this afternoon.

So there is no conspiracy. There is an absence. The signal was always there, the physics was always known, and nobody had put an instrument on the cruising yacht that was intelligent enough to raise its hand the moment the numbers started to drift — rather than a switch that shouts once the argument is already lost.

And before anyone objects on the obvious ground — yes. The senders on your engine are not calibrated instruments. Nobody certified that oil-pressure sender against a reference. It may read a few per cent high, or a few per cent low, and neither you nor we have any way of knowing which. That objection is entirely correct, it is usually where this conversation stops, and it is the reason nobody bothered.

There is one honest limit to that, and it is the same one stated earlier: if the sender itself drifts, that is indistinguishable from the engine drifting. Which is precisely why the first response to any warning is to check the gauge — and why the claim is always “this engine is not making what it used to make”, never “your pump is worn”.

What the Galvanic Voice Does Instead

The Galvanic Voice does not add a sensor to your engine. It reads the four channels your engine is already publishing on NMEA 2000, and it does the one thing nobody was doing with them: it learns what your engine is normally like, and then watches for the deviation.

Step one — the correlation, not the number

Instead of comparing pressure to a constant, the Voice fits your engine’s own relationship between pressure, engine speed and oil temperature. The physics says the form of that relationship: pressure should follow engine speed almost proportionally, and fall with rising temperature as the oil thins. So the Voice fits exactly that shape to your engine’s own data, over thousands of hours, and keeps only the coefficients.

On our own engine, that fit gives an exponent of 0.947 against a theoretical 1.0 — the small shortfall being pump slip and leakage that is not perfectly laminar, both of which push the real exponent slightly below unity. In other words: the physics is confirmed, and the engine’s own numbers are the ones used, not the textbook’s.

Step two — find the valve, and refuse to be fooled by it

Before any of that can mean anything, the Voice has to find the relief valve, because every sample taken while the valve is open is a measurement of a spring rather than of your bearings, and including those samples would corrupt the whole model.

It finds it by measurement, not assumption: it looks for the pressure plateau at the top of the distribution and then proves it is a valve by checking that the pressure stops rising with engine speed across a wide span of revs. A plateau that holds across a factor of three in engine speed is a pressure limiter; a narrow band is just where the boat happens to motor. On our engine the valve sits at 4.76 bar, and every sample on that plateau is excluded from the model that judges the bearings.

Step three — measure how steady the engine actually is

A model is only as useful as your knowledge of its own scatter. An engine at idle is genuinely less steady than an engine at cruise — the revs hunt, the load pulses, the pressure wanders. So the Voice measures the real spread of a healthy engine at each engine speed separately, rather than assuming one figure across the range. On our engine that spread is 2.7 % at idle and about 1 % at cruise — very nearly a factor of three between the two, which is exactly the sort of thing that a single global number gets badly wrong.

The threshold is then set a fixed number of standard deviations below the engine’s own normal, at that engine speed. Not below a constant. Not below the manual. Below your engine, at these revs, at this oil temperature, in the condition it was in when it was healthy.

Step four — the one moment you are needed

Up to here the Voice has done everything by itself. It has been rebuilding its picture of your engine continuously, hour after hour, from nothing but the bus. And then it stops and asks you a question, because there is exactly one thing in this entire method that no instrument can supply.

The question arrives when something makes the old picture obsolete — you have changed the oil, you have had work done, or you have simply decided the conspiracy is real and it is time to start. In the app, under Machinery, the panel is called “Learned oil-pressure baseline”, and it opens by telling you what it is for, in these words:

“Over the hours below, was this engine running normally? No oil-pressure trouble you knew of, no work outstanding on it, nothing you were nursing along.”

That is the whole of your contribution. You are not calibrating anything, you are not entering a number, and you are not being asked to diagnose your engine. You are being asked to state an assumption — during these hours, I believe my engine was working correctly — and the app is explicit that this is what it is, in the confirmation sheet:

And it does not ask you to take any of it on trust. Everything the Voice has worked out is on the same screen, in plain words, with your engine’s numbers in it. This is that panel, as it reads on our own engine:

The labels and the explanatory lines above are the app’s own text, word for word. The values are our engine’s, read off the boat.

Note the second line, because it is the one that makes the instrument work. From the moment you confirm, the Voice stops learning. A model that keeps adapting to your engine will follow a slow decline all the way down and never mention it — which would defeat the only thing we are trying to do. So the picture is sealed, and from then on the Voice measures the difference between the engine you have and the engine you confirmed.

From that moment on, the Voice monitors your engine against its own thermodynamic model — the pressure it should be making, at these revs, at this oil temperature, given what this engine was measured to do — and speaks when it falls away from it.

Which puts a number on the difference. Here is where each instrument opens its mouth, on our own engine, at the speeds you actually motor at:

Read the last column as what it is. At cruise, the difference between the two instruments is three and a half bar of oil pressure — pressure your engine can quietly lose, over a season, with the panel dark and the engine sounding exactly as it always has. The Voice says something after the first fifth of a bar.

And Not One Number in It Came From a Catalogue

This is the part I would most like you to take away, because it is what makes the sensitivity possible rather than merely claimed.

Nothing in this method rests on an assumption about your hardware, and nothing in it is calibrated. There is no lookup table. There is no “typical marine diesel”. There is no figure taken from a workshop manual — we do not have your engine’s workshop manual, and we do not need it. Every quantity in the model is a measurement of the actual object sitting in your boat:

• The pressure–speed law is your engine’s. The exponent came out at 0.947 on ours. Yours will be slightly different, because your pump, your clearances and your two thousand hours are different. Whatever it is, that is the number that gets used.
• The relief valve is your valve. 4.76 bar on our engine — found by measurement and then proved to be a valve by watching the pressure stop rising with revs across a wide span. Not a catalogue figure, not a design intent, not a spring rate from a parts diagram. The one in your engine, as it is today, after however many hours and whatever has happened to it.
• The scatter is your engine’s scatter, measured separately at each speed band — because a fast idle on your installation genuinely wanders differently from ours.
• The oil is whatever is actually in your sump. Not the grade printed on the bottle. Not the grade in the service schedule. The fluid that is in there right now — including, if we are all being honest with each other, the litre of something-not-quite-the-same you topped it up with in a marina in August because it was what the chandler had. The model never asks what the label said. It measures what the fluid does.
• The temperatures are the ones your thermostat actually holds — 79.9 to 82.9 °C on ours, measured, not the 80 °C somebody wrote in a specification.
• Even the alarm threshold is not chosen. It is set a fixed number of standard deviations below your engine’s own normal, where those standard deviations are themselves measured from your engine. Nobody picked a round number and nobody tuned it until it felt right.

The only human input in the whole chain is the single sentence in Step Four — and that is an assumption about history, which you are the only one who can supply, clearly marked as such, and honest about its consequences if it is wrong.

Which is the whole purpose of the exercise. Not to be alarming, and not to be clever — but to notice the first small departure from what your engine, with your oil, through your valve, was doing when it was well, and to say so while there is still nothing to repair but a gasket, a cooler or a filter. The point is not to announce the failure. It is to keep it from ever happening.

By How Much Does This Actually Improve Things?

Everything below is measured. Not modelled, not estimated, not scaled from a brochure. It comes from 543,755 engine records read off the NMEA 2000 bus of our own boat — a Hanse 588 with a Volvo Penta D3-110 at 2,098 engine hours — over a week of real cruising in August 2026.

That engine’s own learned normals: 2.235 bar at hot idle (about 780 rpm), 4.154 bar at cruise (about 1,500 rpm), with the oil settling at 80–83 °C and the relief valve measured at 4.76 bar.

The honest way to compare instruments is to ask: how much of the pressure has already been lost before this instrument says anything?

Put as multiples, which is the number worth remembering:

And now the earlier point about delivery pays off. A ten per cent loss of gallery pressure does not mean the engine is running at ninety per cent of its lubrication. It means the rocker gear, or the piston jets, or the turbo — whichever point the designer left with the least margin — has begun to go short, while every other bearing in the engine is still perfectly fed and nothing whatsoever sounds wrong. Two parts in a thousand is not an absurdly small thing to look for. It is the correct place to look, because that is where the countdown starts.

Three orders of magnitude is not a marketing number. It is what happens when you stop judging one reading against a constant and start averaging thousands of readings against a measurement of the same engine when it was well.

And one last figure, which is the one I find most convincing, because it is the one that could have embarrassed us. Our engine’s actual measured drift, right now, at 2,098 hours, is −0.124 % — which against the noise of the measurement is 1.7 standard deviations, i.e. below the detection threshold. The instrument looked as hard as it can look, and reported that there is nothing there.

That is what a real instrument does. It does not find a problem to justify itself.

What We Are Not Claiming

Three things, stated plainly, because a piece full of numbers has an obligation to say where they stop.

We cannot tell a failing sender from a failing pump. The honest claim is “this engine is not making what it used to make at these revs and this oil temperature”. The first thing to check is always the gauge.

We cannot see oil level. That is still the dipstick’s job, and the dipstick is still your friend. Keep wiping it on the terrible rag.

We cannot yet turn a percentage into a number of months. To say “you have four months” we would need to know how fast your engine loses pressure, and that requires watching engines that are actually declining, over seasons, on more than one boat. We are measuring it. We are not going to invent it. What we can say is a ratio, and the ratio is the whole argument: the switch is looking at the end of the decline. We are looking at the first few per cent of it.

So, a Conspiracy?

No. Something more ordinary, and in a way more irritating.

Every ingredient of a proper engine-condition monitor has been sitting on your boat for years. The physics was settled by Reynolds in 1886. The engine has been designed against known parameters since Diesel. The four sensors are already fitted and already talking on the bus, twice a second, every second you motor — and, as we have seen, they do not even need to be calibrated for this to work. Industry has trended exactly these numbers since the 1970s.

The missing piece was never a sensor, a standard or a discovery. It was something on board patient enough to learn what your engine is like when it is well, and attentive enough to notice, quietly and early, when it stops being that. Not a louder alarm. Not another screen. An instrument that raises its hand at two parts in a thousand, and says so out loud, while the repair is still a gasket and an afternoon.

And to have that on your boat, you install a Galvanic Voice. That is the entire prerequisite — that, and an engine already connected to NMEA 2000, which is essentially every marine diesel sold in the last twenty years and a great many older ones behind a gateway. There is no work on the engine itself. Nothing is tapped, nothing is drilled, no sender is replaced, no probe goes into your oil. The engine is already saying all of this out loud on the bus; the Voice simply listens, learns your engine over your hours, and speaks.

There is no subscription either — no monthly fee, no tier, no cloud plan that quietly expires and takes the feature with it. You buy the device and the device is yours, working, on your boat.

Nor does it only watch the engine. The same unit is listening to everything else the bus carries and everything it can sense for itself — where the anchor is and whether it is holding, who is on board and whether they are still on board, what is converging on you and how long you have, what the batteries and the tanks and the bilge are doing, what the weather is about to do — and it says all of it out loud, in the language you sail in, at the moment it matters rather than in a menu you would have had to open.

And the arithmetic of that is worth a sentence on its own. A Galvanic Voice costs €950 ex-VAT — and it is worth being clear about what that buys, because everything in this article is one of the things it does, not the thing it is. The engine watch is a single item on the list above. The same box, for the same money, is simultaneously minding the anchor, the crew, the traffic, the batteries, the tanks, the bilge and the weather, and speaking about all of them. The oil-pressure model is, so to speak, a by-product of already being aboard and already listening.

Against that: a new Volvo Penta D3-110 — the engine it is watching over — lists at a UK Volvo Penta dealer at £19,129 including VAT, and that is the bare engine: no saildrive, no gearbox, no installation, nobody paid to fit it. Call it twenty Galvanic Voices for one engine, and that is the flattering version of the comparison.

Then add the saildrive. Then the yard’s labour. Then the fact that on this particular boat the job appears to begin with a saw and end with a new galley entrance. Then the lost season, the tow, and the afternoon off St Vincent. At some point the ratio stops being interesting and starts being slightly embarrassing.

So Who Else Does This?

It is a fair question and we went looking for the answer properly, because a claim like the one we are about to make deserves to have been checked rather than assumed.

Here is the whole of what exists for a pleasure boat today, in four kinds:

• Gauges. Digital instruments that take the oil-pressure number off the bus and put it on a screen, prettily. A better-looking needle. No analysis of any kind.
• Monitoring and telematics boxes. Remote alerting, historical charts, and thresholds you set by hand — you choose a number, and it tells you when the number is crossed. Which puts you back where we started: what number should a person who is not an engine designer type into that box, for an engine whose correct pressure changes with every rev and every degree?
• Usage-based service reminders. Engine-hour counters that tell you a service is due. That is a calendar, not a condition. It knows how long the engine has run; it knows nothing about how it is running.
• The engine makers’ own apps. A dashboard, trip data, and alerts on the engine’s diagnostic trouble codes. Which is to say: the same switch, forwarded to your phone. The app makes the death certificate mobile. It does not make it arrive any earlier.

What we could not find — anywhere, in any product sold to the owner of a sailing yacht or a motor cruiser — is a system that fits your engine’s own pressure law from your engine’s own data, normalises it against both engine speed and oil temperature, finds your relief valve by measurement and refuses to be fooled by it, sets its threshold from your engine’s own measured scatter at each engine speed, and then freezes that picture and reports the drift away from it.

And Where the Method Is Standard Practice — Which Is Rather the Point

We should be equally clear about what is not new here, because it is the most reassuring part of the whole story.

The mathematics is not our invention. Trending a machine against a model of its own normal behaviour is mature, conservative, thoroughly proven engineering — and it is exactly what is done in the places where an engine failure is not an inconvenience but an emergency:

• Commercial shipping and industrial plant, where condition monitoring and lubricant trending have been routine since the 1970s, and where a main engine is watched continuously against its own history rather than against a red line.
• General aviation. Piston-aircraft owners have had engine trend monitoring for years — data logged every flight, compared against that aircraft’s own previous behaviour under similar operating conditions, with statistically significant shifts in parameters like falling oil pressure flagged long before anything is out of limits.

We regard that as the best possible endorsement of the approach. The company this method keeps is aviation and commercial shipping. When an aeroplane engine and a 30,000-tonne ship are both watched this way, and a $1M boat is watched by a pressure switch from 1955, the question is not whether the method works. The question is why nobody had bothered to bring it down to the size of our boats.

Which is also why we are comfortable saying the next part out loud: we think it will be difficult to do this better. Not because we are clever — the physics did the hard part in 1886 and the engine designers did the rest — but because there is very little room left between what the sender can resolve and what the engine’s own scatter allows. We are already measuring two parts in a thousand on an instrument quantised at four hundredths of a bar. Anyone who wants to improve on that will have to argue with the sender, not with us.

A Closing Word on the Conspiracy

We should probably declare our position. We believe in science. It is an unfashionable credo, it makes for very poor conversation at anchor, and it has one enormous practical advantage: it renders conspiracies entirely irrelevant to us.

Consider what this one would actually require. Every engine manufacturer on earth, across a dozen countries and a century of ferocious competition, quietly agreeing to fit exactly the right four sensors, to publish them on an open standard twice a second for anyone to read — and then, with iron discipline, generation after generation, declining to do the arithmetic. The minutes of that meeting would be the finest document in maritime history.

And here is the cheerful part: it does not matter which way it goes. If the conspiracy is a fantasy, then this was simply an idea nobody got round to, and now somebody has. If the conspiracy is real — then we have, entirely by accident, published the antidote, in full — and it fits every boat whose engine is on NMEA 2000, which is very nearly all of them. Either way you end up in the same place. That is the great charm of arithmetic: it does not require anybody’s cooperation.

As for whether ours is the best way to do this — we think it is, and we have taken the faintly reckless step of making that easy to check. It is all in this article. The model. The exponent. The valve, and how we prove it is a valve rather than a coincidence. The scatter at every engine speed. Where the threshold sits and why it sits there. What happens if you answer the confirmation question wrongly. And a list of three things we cannot do at all. We even published the measurement that found nothing — our own engine’s −0.124 %, sitting below its own detection threshold, stubbornly refusing to justify the instrument watching it.

None of that is generosity. It is the cheapest form of confidence there is: you do not publish your method if you are nervous about somebody reading it. We would warmly encourage you to go and look for the equivalent document from anybody else.

We will wait here, humming quietly — the way an engine hums when it is genuinely well: every parameter sitting inside its own window, nothing drifting, nothing hidden. In no pain at all.

Your engine has been telling you how it feels for its entire life. Nobody was listening carefully enough to hear it.

Give a voice to your engine, too.

References

1. Tower, B. “First Report on Friction Experiments.” Proceedings of the Institution of Mechanical Engineers, 1883. (The railway-bearing experiments that accidentally discovered hydrodynamic lubrication — the reason a steel journal never touches its shell.)
2. Reynolds, O. “On the Theory of Lubrication and its Application to Mr. Beauchamp Tower’s Experiments.” Philosophical Transactions of the Royal Society, 1886. (The governing equation of the oil film in every plain bearing in your engine.)
3. Stribeck, R. “Die wesentlichen Eigenschaften der Gleit- und Rollenlager.” Zeitschrift des VDI, 1902. (The friction curve that separates hydrodynamic, mixed and boundary lubrication — the map of how a bearing fails as the film thins.)
4. Diesel, R. German Patent DE 67207, 1892; first working engine, 1897. (The thermodynamic idea, essentially unchanged since.)
5. Cantley, R.E. “The Effect of Water in Lubricating Oil on Bearing Fatigue Life.” ASLE Transactions, 20(3), 1977. (The classical measurement of how little water contamination it takes to matter — rolling-element bearings, but the mechanism generalises.)
6. Hamrock, B.J., Schmid, S.R. & Jacobson, B.O. Fundamentals of Fluid Film Lubrication. 2nd ed., CRC Press, 2004. (Standard modern treatment of film thickness, viscosity and load capacity.)
7. Heywood, J.B. Internal Combustion Engine Fundamentals. 2nd ed., McGraw-Hill, 2018. (The reference text on engine lubrication circuits, pump sizing and relief-valve behaviour.)
8. Taylor, C.M. (ed.) Engine Tribology. Elsevier Tribology Series, 1993. (Bearing, ring and valve-train lubrication in reciprocating engines specifically.)
9. ASTM D6224 — Standard Practice for In-Service Monitoring of Lubricating Oil for Auxiliary Power Plant Equipment. (The industrial condition-monitoring practice that has existed for decades and never reached the cruising yacht.)
10. Volvo Penta D3-110 list price, French Marine Motors Ltd (UK Volvo Penta dealer), retrieved August 2026: from £19,129.20 inc VAT, engine only, saildrive and gearbox not included. frenchmarine.com
11. NMEA 2000 PGN 127488 (Engine Parameters, Rapid Update) and PGN 127489 (Engine Parameters, Dynamic). (The two standard messages carrying engine speed, oil pressure, oil temperature and coolant temperature on the boat you already own.)