In the event that someone reads and understands my short description of sailing around the world with a 35 year old Yanmar SB8 for power it would be appropriate to add a few words about diesel engines in general. The first and most important note is that although 0.15GPH seems like great fuel economy for an ocean going vessel, the Yanmar SB8 is not all that efficient of an engine. A real diesel engine would use hardly more than half as much fuel to do the same job of pushing Eva along at three and a half to four knots. That said the 0.15GPH is probably quite a bit better than anyone could get with an off the shelf marine diesel engine from the '80s, '90s or the first decade of the 21st century for that matter. What is going on is that although the Yanmar SB8 is a rather inefficient and dirty old engine of the indirect injection combustion system type it does have a regulator valve type injection pump that does a fairly good job of delivering the fuel at the correct time and at the correct injection flow rate for a reasonably wide range of speeds and loads. Nearly all marine diesel engines of less than about 50hp from the past thirty years have used Bosch type inline injection pumps with metering collars that can deliver the correct timing and injection flow rate over a disappointingly narrow range of speeds and loads. For more on the few direct injection sailboat auxiliary engines that have been available over the years see Small Marine Diesel Engines.
Why do indirect injection diesel engines exist?
Where have all the regulator valves gone?
Metering Collar Switcharoo
The Common Rail Solution
Engine Speed
Specific IDI Problems
Conclusion
There is no good answer to this question, but a few notions have been suggested over the years. The most reasonable explanation is that indirect injection (IDI) diesel engines came about due to inadequacies in injection systems. Pre-combustion chambers can control the combustion process in a number of ways that are desirable for hiding the inadequacies mostly of the Bosch type inline injection pump. A really good indirect injection diesel engine sounds like it is running good even if it is in fact burning a huge amount of fuel and wearing out quickly. This is good for selling engines, but is horrible for most other purposes. This explanation somewhat breaks down though in light of the existence of the rotary distributor type injection pump with a regulator valve that can do an excellent job of delivering the fuel at the correct time and at the correct injection flow rate for nearly any application. A direct injection engine with a rotary distributor injection pump can sound like it is running well while actually delivering high efficiency and spectacular longevity.
As with many marvels of the modern world the rotary distributor injection pump (also called a radial distributor injection pump) mysteriously showed up in the early 1950's. A diesel mechanic and engineer named Vernon Roosa is given credit for inventing the rotary distributor injection pump which was marketed by the Stanadyne Corporation and was apparently very popular for some time. Going back to at least the 1930's and probably all the way back to Rudolph's original high pressure air injection systems single cylinder diesel engines worked quite well, but there was no good way to get multiple cylinder engines to stay in balance. The solution was the Bosch inline type injection pump that used metering collars which stayed in balance when used on multiple cylinder engines. The only problem with the metering collar type injection systems was that they were so inflexible as to cause the diesel engines to work only over a very narrow range of speeds and loads. Since larger power output and better transmission performance required multiple cylinder engines the regulator valve was mostly left behind. Left behind by everyone but Vernon Roosa who made a living in the 1930's building injection systems for large stationary generator engines. Probably in trying to get two regulator valves to work on two cylinder engines Roosa came up with a way to inject fuel into more than one cylinder with just one regulator valve. With the support of the Stanadyne Corporation, who bought his idea and hired him to develop it into a marketable product, the rotary distributor injection pump was born. Why the rotary distributor injection pump has fallen out of use to the point where it has not been possible to buy a new one in decades is something of a mystery. The largest potential problem with the rotary distributor injection pumps is that inevitable wear of the regulator valve over time would tend to throw fancy governor mechanisms out of adjustment. What this means is that a really reliable regulator valve type injection system is able to inject more fuel than the engine can use. If other mechanisms are not used to limit the maximum fuel delivery engine overloading may occur. On a normally aspirated engine visible black smoke is an indicator that the engine is being run too hard, and on a turbocharged engine a certain boost versus engine speed curve indicates the safe maximum load. When maximum boost is capped at high engine speed by a waste gate then there is no indication of engine overloading other than how much power it is putting out, which may not be sufficient to limit fuel delivery before the pistons melt. On a boat it is fairly easy to limit maximum power delivery by measuring fuel consumption at various engine speeds to determine how fast the prop can be turned. For on road vehicles it requires a bit more of a seat of the pants approach to use a rotary distributor injection pump, which tends to be problematic. Another very significant problem that many regulator valve equipped engines have had is the lack of a fuel cutoff port for ending the injection event. If the injection flow rate is allowed to slowly drop off at the end then dribbling of the last bit of fuel tends to cause smoky and smelly operation. Internal passages in plungers to allow crisp termination of the injection event is no more of a technical challenge for regulator valve based injection systems than the internal plunger passages required for all metering collar based injection systems.
It is usually considered the case that Vernon Roosa's rotary distributor injection pump was the first application of a single regulator valve to a multiple cylinder engine, but this may in fact not be totally true. According to J. Herbert Wickman in his 1935 book "The DIESEL GUIDE" the Ingersoll-Rand Company used Price brand linear distributor injection pumps with regulator valves instead of metering collars (pp. 36 "...the plunger makes a separate stroke for each cylinder...a bypass valve regulates the measure of fuel supplied...) Here Wickman calls the regulator valve a "bypass" valve, and that is exactly what it does. The regulator valve bleeds fuel off of the high pressure circuit to reduce the injection flow rate. Usually it is said that the advantage of rotary distributor injection pumps is that they can handle higher pressures and are more reliable at higher engine speeds on high cylinder count engines. As the Robert Bosch Co. and Volkswagen demonstrated in the late 1970's linear distributor injection pumps can however hold up reasonably well on four, five and even six cylinder engines up to 4500RPM. Rotary distributor injection pumps probably are mechanically more robust, but ultimately the major distinction is that rotary distributor injection pumps can not use metering collars. A linear distributor injection pump can use either a regulator valve or a metering collar, although modern examples seem to always use a metering collar instead of a regulator valve.
What metering collars are is absolutely consistent and reliable. That sounds really good for a diesel engine, too bad they only work over a narrow range of engine speeds and loads. There are two different types of metering collars, the axial metering collar and the rotational metering collar, but functionally they are exactly the same. The axial type metering collar slides up and down on the plunger to uncover the fuel cutoff port in the plunger at the selected end of injection timing value. The rotational metering collar has an angled slot in the side and is rotated so that the slot uncovers the fuel cutoff port in the plunger at the selected end of injection timing value. The insurmountable limitation of metering collar control of fuel delivery is that the injection flow rate remains constant at all engine loads. On inline and unit injector type pumps with a seperate metering collar for each cylinder the main limitation is that the injection start timing ends up far too early at lower engine speeds causing loud, harsh and extremely inefficient operation under light loads. The linear distributor type injection pumpes (also called axial distributor injection pumps) which use a single plunger and a single metering collar and incorporate a speed dependant advance mechanism do go a long way to getting an engine to run over a wide range of speeds. This is highly desirable, and even useful for some applications. The fixed injection flow rate is however still somewhat of a problem. The fixed injection flow rate not only means that the range of loads under which the engine can operate efficiently is severely limited but also that the range of speeds over which the engine can operate is somewhat limited. The reason for this is that lower speed operation would tend to require longer injection duration than high speed operation. A linear distributor injection pump with a metering collar delivers just the opposite of this, longer injection duration at high speed and shorter injection duration at low engine speed. The thing is that a linear distributor type injection pump with a metering collar works so much better than a Bosch type inline pump just by virtue of having a speed dependant advance mechanism that other problems tend to get overlooked.
The injection flow rate problems can be partially solved with a fancy system of multiple fuel cut off ports that allow a lower injection flow rate at lower power output. This system probably works fairly well, and is in fact the only viable alternative to a regulator valve for a mechanical injection system. The limitation of these multiple cut off port injection flow rate varying systems is that the flow restriction on the auxiliary fuel cut off circuit has to be carefully matched to the flow characteristics of the high pressure circuit, including the injectors, for the reduction in injection flow rate to work over a wide range of engine speeds. If there is any mismatch between the flow restriction and the flow characteristics of the high pressure circuit then the reduced injection flow rate will only be matched to the flow rate required by the engine for low power output at one engine speed. This type of injection flow rate reduction system would also tend to be unusually sensitive to changes in fuel viscosity. Even when an auxiliary fuel cut off port system is working perfectly it is of limited utility since the transition from low injection flow rate to high injection flow rate causes a situation where the injection flow rate is high early in the injection event and then drops to the lower injection flow rate for the later part of the injection event. This is just the opposite of what would be ideal to get a diesel engine to run well, and the auxiliary fuel cut off port system for reducing the injection flow rate ends up being useful mostly just for getting the engine to idle smoothly and quietly at low engine speed with little or no load on it.
As far as the injection start timing goes the linear distributor metering collar type injection pump with it's speed dependant advance mechanism and cold start advance mechanism can do a quite good job of getting it perfect. The problems with the inline type injection pump mostly center around lag in the high pressure injection lines. The longer the lines the more lag time there is and the worse the timing problems tend to be. An elongated fuel inlet port shape can provide some advancing effect, and on a good inline injected engine the start timing can be pretty close to correct over a certain range of engine speeds. A unit injection system is essentially the same as an inline type pump, but the high pressure lines are of course very short. Unit injector systems can make very good use of the advancing effect of an elongated fuel inlet port, and injection start timing can come out fairly close to correct over an even wider range of engine speeds. A severe limitation though on both inline pumps and unit injection systems is that there is no way to provide a cold start advance other than setting the static timing so the engine will start and run in cold conditions. What this typically means is that even the best inline injected or unit injection engines have far too early of injection start timing at low idle. Metering collar type linear distributor injection pumps can do much better, both in terms of low idle timing and in terms of perfect start timing over a wide range of operational speeds.
In the end though the worst problem with all of these metering collar based injection systems is that the fixed injection flow rate means that the engine only works well over a limited range of loads. If the injection flow rate is set for high power output then radically reduced power output is loud, harsh and very inefficient. If the injection flow rate is set lower then low power output operation improves, but the engine is then not able to make as much power and efficiency suffers under heavier loads. To make matters worse even under a heavy load lower engine speed operation would tend to require a lower injection flow rate and higher engine speed would tend to require a higher injection flow rate. This is why metering collar equipped engines tend to be able to make monsterous amounts of torque at lower engine speeds and a disappontingly low maximum power output at higher engine speeds. For on road vehicles the huge torque and a flat power output curve works just fine, as higher engine speed is only required to get into the next gear. For a marine engine though maximum power output is required at maximum engine speed, and the load drops of dramatically at reduced engine speed.
Computer controlled linear distributor injection pumps and computer controlled rotary distributor injection pumps have both been used over the years. These electronically controlled injection pumps do essentially the same thing as the fully mechanical metering collar type linear distributor injection pump. That is full control of injection start and end timing is possible at any engine speed but the injection flow rate is still fixed. The main difference is that the electronically controlled injection systems are mechanically simpler and are able to provide absolutely perfect control of start and end timing at any engine speed and load. The computer controlled rotary distributor injection pumps are a bit confusing in that they use an electronically controlled "metering collar emulator" instead of a regulator valve. The fuel cutoff port still controls the start of injection timing and the electronically controlled valve takes the place of the metering collar in controlling the end of injection timing. With the use of an electronically controlled actuator on the advance mechanism to rotate the cam ring full electronic control of both start and end timing is provided for, but the injection flow rate remains fixed just as on a fully mechanical linear distributor injection pump. It is only the injection pumps that use a regulator valve which are able to match the injection flow rate to the requirements of the engine over a wide range of speeds and loads. It is important to keep in mind that an electronically controlled regulator valve would be technically feasible and could be installed on any type of injection system to allow full control of the injection flow rate at any engine speed or load.
Electronically controlled injection systems can also easily incorporate small adjustments for engine temperature, ambient air temperature and altitude although these have not been widely used. Any turbocharged engine with a distributor type injection pump can have a boost dependant timing mechanism, but due to fixed injection flow rate limitations the boost dependant timing mechanisms normally are of only limited use on linear distributor injection pumps. The reason that boost dependant timing mechanisms are of essentially no use with a fixed injection flow rate is that at higher engine speeds and higher loads the injection durration is so long that making the injecttion end timing later would cause smoky operation with the last of the fuel injected after the cylinder pressure has dropped off to the point that it can not burn completly. With a high injection flow rate for a very high output turbocharged engine a boost dependant timing mechanism would however be of some use, particularly at the lowest engine speeds where substantial boost is available. On regulator valve equipped rotary distributor injection pumps a boost dependent timing mechanism is of great benefit as the injection flow rate increases under heavier loads preventing the injection end timing from becoming too late. A bit of cruel irony here is that the electronically controlled injection systems can very easily incorporate a boost dependant timing mechanism, but it is only the fully mechanically controlled rotary distributor injection pumps with regulator valves that really benefit from boost dependant timing mechanisms.
It should also be mentioned that a rotary distributor injection pump with a boost dependant timing mechanism also needs to have a speed dependant timing mechanism to compensate for lag in the high pressure lines. A rotary distributor injection pump on a normally aspirated engine can get away with no speed dependant timing mechanism at all, although this does lead to an engine that runs best under a heavy load at high engine speed and runs best under a light load at low engine speed. On any engine, normally aspirated or turbocharged, a rotary distributor injection pump works best with both a speed dependant timing mechanism and a load dependant timing mechanism, and a load dependant timing mechanism absolutely cannot be used without a speed dependant timing mechanism (A boost dependant timing mechanism is of course a type of load dependant timing mechanism). The classic example of this was the 1980's Ford light trucks with the normally aspirated 6.9 liter Navistar V8 IDI pre-chamber diesel. The 6.9 Navistar used a regulator valve equipped rotary distributor injection pump that had a load dependant timing mechanism with no speed dependant timing mechanism, and usually worked a whole lot better if the load dependant timing mechanism was simply disconnected.
Most diesel engines from the 1970's, through the first decade of the 21st century have used either a Bosch type inline pump or the very similar unit injector system, and many of them have appeared to work quite well. The main reason that these crude diesel engines seem to work so well is that what they are compared to are pathetic gasoline engines designed and installed to run at 2,000 to 4,000RPM, which is too slow for gasoline engines. Another reason that many of these crude inline injected engines appear to work quite well is that they use exhaust gas turbine driven turbochargers that improve the performance of inline injected engines in two separate ways. The most severe limitation of inline injected engines is typically that the maximum engine speed is capped by late injection start timing due to lag in the high pressure lines. With two or even three atmospheres of boost the compression pressure in the cylinders is considerably higher which allows the fuel to light off well even with much later injection start timing. The larger amount of air in the cylinders and the larger quantity of fuel burned also keeps the cylinder pressure high longer, allowing fuel to be burned somewhat later than would be possible for a normally aspirated engine. The other way that a turbocharger gets an inline injected engine to work better at elevated engine speeds is through co-generation. The last of the fuel injected late at high engine speed and high power output tends to burn slowly and incompletely since the cylinder pressure and temperature are dropping off rapidly at that point. As the last of the combustion byproducts continue to react with excess oxygen in the exhaust, heat continues to be generated long after the exhaust valves open. On a normally aspirated engine all of this slowly reacting after combustion is wasted. On a turbocharged engine the waste heat of the engine and much of the slowly reacting after combustion is used to drive the intake air compressor. Not only does turbocharging increase the amount of fuel that the engine can burn, but it also does some useful work of driving the pistons down on the intake stroke. A 15 liter four stroke engine running at 2000RPM might be able to make 300 or 350hp normally aspirated and upwards of 1000hp turbocharged. If the intake air for this 15 liter engine is being driven in at three atmospheres of boost then the pistons being forced down on the intake stroke are going to do something like 100 or 120hp of output at 2000RPM. In a very real sense a turbocharged engine can begin to look more like an internal combustion turbine engine and a steam engine than a reciprocating piston internal combustion engine, and this also tends to mask problems with the injection system.
A computer controlled common rail injection system should be able to vary the injection flow rate based on engine speed and load, but there is some evidence that common rail systems in common use do not in fact very effectively reduce the injection flow rate for reduced power output. What most common rail injection systems do however do at low speed is use one or two small early injections of fuel to keep the combustion chamber hot enough to deliver the main charge of fuel later when it can more efficiently be used to drive the crankshaft. This can be quite effective in keeping the engine quiet at low engine speed and is generally a good feature for efficient operation. Getting the main charge of fuel to burn latter when optimal conversion efficiency at the crankshaft is possible would tend to keep efficiency high at low engine speed. If however the injection flow rate is too high for low speed operation then loads on engine bearings will be unnecessarily high which will cause the engine to be less efficient under all conditions and wear out more quickly. A computer controlled common rail injection system should very easily be able to deliver just the right injection flow rate at just the right time for extremely efficient operation over a very wide range of speeds and loads.
Regulator valve type mechanical injection systems have the very desirable characteristic of having an injection flow rate that starts off low and then gets higher towards the middle of the injection event. This is very useful for low power output operation because it tends to stretch the pressure peak around to where the conversion efficiency is highest. At higher engine speeds regulator valve type mechanical injection systems use the fuel inlet port to begin the start of injection more crisply which allows enough fuel to be dumped in early enough for maximum power output. The obvious problem with this is that overfueling at low engine speed would cause not only more fuel than could be burned to enter the combustion chamber but also for that fuel to be injected too fast and too early which puts huge loads on the engine. Combined with difficulties in keeping sophisticated governor mechanisms adjusted on regulator valve type mechanical injection systems this limitation was undoubtably part of the reason that the rotary distributor injection pumps were not more widely used. It is not that regulator valve type injection systems cannot work perfectly over a very wide range of speeds and loads, it is just that they were not always idiot proof. The high loads encountered by accidental overfueling meant that large, heavy, overbuilt diesel engines held up best, and tended to last darn near forever if they were never over fueled.
A computer controlled common rail injection system could be made that would work perfectly under all conditions, and this could be accomplished in a number of different ways. Adding an electric motor driven regulator valve to bleed off excess fuel at low engine power output would be a simple and effective way to accomplish this, but a variable speed electric motor driven high pressure pump or multi stage injectors could also be used to accomplish the same thing. Modulating the pressure in the high pressure circuit to adjust the injection flow rate would be as good as any regulator valve type mechanical injection system with two distinct and very significant advantages. Full electronic governing of fuel delivery would mean that there would be no overfueling problems and engines could be lighter and last longer. The small pre-injection squirt of fuel used to keep the combustion chamber hot during low speed operation would allow higher efficiency while also preventing extremely low injection flow rates that would cause injectors to dribble instead of spray. Even with fully modulatable rail pressure a common rail injection system could benefit from multistage injectors that would spray nicely at lower injection flow rates. Regulator valve type mechanical injection systems would also have benefited from multistage injectors for the same reason.
Ideally the injection flow rate profile, that is how the injection flow rate builds throughout the injection event, would also change to match the exact requirements of the engine at any particular engine speed or load. Regulator valve equipped rotary distributor injection pumps can sort of do this in that a well shaped fuel cam delivers more taper to the injection flow rate at low power output and delivers substantially less taper at higher power output. An electronically controlled regulator valve could also quite easily provide at least this basic level of injection flow rate profile to power output matching, and more sophisticated injection flow rate profile modulation would also be possible. The ultimate sophistication for fully mechanical injection systems would be a linear distributor injection pump with both a regulator valve and a metering collar. Adding the metering collar would allow different parts of the fuel cam profile to be used for different conditions, although controlling all of this purely mechanically and hydraulically would make for quite a Rube Goldberg contraption. A computer controlled injection system with an electronically controlled regulator valve and full electronic control of both start and end timing could much more easily shift the injection event back and forth on the fuel cam in order to match the injection flow rate profile to the requirements of the engine for each particular speed and load combination. The important thing to keep in mind is that even with a mostly flat injection flow rate profile extremely good results can be attained. The biggies are getting the start timing correct for each engine speed and load combination as well as being able to reduce the injection flow rate for partial load operation.
Shifting the injection event back and forth on the fuel cam to provide injection flow rate profile matching for different engine speeds and loads could potentially eliminate the need for an electronically controlled regulator valve for varying the injection flow rate. The reason to eliminate the regulator valve is to reduce pump driving power. Because the same volume of fuel must be pressurized for light load operation as for full load operation it might be expected that a regulator valve type injection pump would require a lot more power to drive. The reality though is that the pressure reduction accounts for a large amount of the difference, and in practice regulator valve type pumps do not require all that much more drive power. The fact that the pump drive power normally remains quite low compared to engine output means that the use of a regulator valve is not a real hindrance to efficient operation. Still though there is some advantage to not using a regulator valve, and a different means of reducing injection flow rates for light load operation might be considered desirable. The main obstacle to an electronically controlled injection system that is able to shift the injection event back and forth on the fuel cam is that it requires an electronically controlled valve that is capable of closing under pressure. The solenoid valves used on electronically controlled distributor type injection pumps are only required to open under pressure, and then they close while the high pressure circuit is not pressurized. An electronically controlled valve that can also close under pressure would require a linear electric motor instead of a simple solenoid. This would be possible to implement, but the control circuitry required to get it to work is significantly more complex. As a side note these linear electric motors are the same sort of devices required to get fully variable intake and exhaust valve timing systems to work really well.
It is also important to bear in mind that the range of speeds that a particular engine can efficiently run over is ultimately limited by the combustion properties of the fuel, even if the injection system is up to the task of varrying the injection flow rate and timing values to match the requirments of the engine (See section on
Fuel Properties). There are two different concepts relating to speed limitations of engines, these are engine speed and piston speed. Generaly diesel engines can be made to run over a very wide range of engine speeds, provided that the piston speed is not radically high or radically low. At very slow engine speeds of less than about 700RPM diesel engines don't run quite as well because the cylinder pressure drops off to the point where fuel can no longer be burned before the crankshaft has advanced to the point where good conversion efficiency can take place. This is the same reason that gasoline engines do not run well at less than about 4,000 or 6,000RPM. Because the prolonged injection of fuel increases the time of combustion diesel engines can run down to much lower speeds, and all the way down to 150RPM diesel engines have been able to run decently efficiently. At very low engine speeds higher compression ratios work better because the higher compression pressure allows the start of injection to be farther past top dead center. At the other end diesel engines usually don't run well at more than about 4,000RPM because the fuel would have to be injected extremely fast for it to all burn before the cylinder pressure drops off too much. With a sufficiently high injection flow rate diesel engines can be made to run up to 5,000 or even 6,000RPM and even higher speeds would be possible. At very high speeds though diesel engines begin to run into the problem of the fuel not being able to mix with the air fast enough, and the maximum engine speed for a diesel engine would always be lower than for a gasoline engine. At very high engine speeds lower compression ratios work better because the fuel has to be injected earlier to have a chance of mixing with the air in time for it to burn.
Within this range of speeds for a diesel engine of 700 to 4,000RPM for maximum efficiency and 150RPM to 6,000RPM for decent efficiency there is also the piston speed limtation which has to do with the stroke of the engine and on the top end has to do with the weight of the pistons and rods. A normally aspirated gasoline engine with a stroke of four inches can be made to run at 7,000RPM with the use of extremely lightweight pistons and rods and a two inch stroke gasoline engine can likewise run at up to 14,000RPM. Generally the easiest way to attain very high piston speeds with a gasoline engine is with the use of a radically oversquare configuration where the bore diameter is large compared to the stroke length. The large bore makes room for large valves and large ports in the head to allow good flow at high engine speeds. Forced induction can also be used to attain even higher piston speeds, and four inch stroke and four inch bore forced induction racing engines are routinely run at 8,000RPM or higher. For diesel engines piston speeds are kept much lower for good light load efficiency and long engine life. A six inch stroke diesel engine would typically run at about 1000 to 2200RPM with somewhat higher speeds sometimes used for high output turbocharged engines. A three inch stroke diesel engine can easily run up to 4,000RPM or higher, but speeds less than about 1500 or 2000RPM begin to run into a severe problem of insufficient piston speed. The three inch stroke diesel engine can actually still run quite well at speeds less than 1500RPM provided the injection system can deliver the fuel at the correct time and at the correct injection flow rate, but the maximum atainable efficiency does drop off. Rather conveniently slightly insufficient piston speeds are not as much of a problem during light load operation, so big diesel engines can be idled down to quite low speeds with fairly good results. The point here is that although the combustion properties of the fuel do ultimately limit the range of speeds that an engine can run over; quite good efficiency can be attained over a reasonably wide range of engine speeds provided that the injection system is up to the task of delivering the fuel at the proper time and at the correct injection flow rate.
There are two main problems that the indirect injection combustion system causes, these are excess heat transfer to the cooling jacket under a light load and a flow restriction that limits maximum power output. On just about all IDI engines it is easy to observe that they use a whole lot more fuel at very low power output than direct injection engines. An IDI engine of one liter displacement and rated at 25hp with an inline injection pump might burn a quart an hour at low idle, where a direct injection engine of six liters displacement and rated at 120hp with an inline injection pump would also burn about a quart an hour at low idle. That is a really very dramatic difference, and part of the explanation for that difference is that at low power output all of the fuel burned by the IDI engine is burned up in the large in-head pre-combustion chamber. With all of the fuel burning so far up into the cylinder head and so far away from the piston that is supposed to be pushed down by that combustion much more of the heat of combustion is transferred to the cooling jacket. The fact that the large in-head pre-combustion chamber causes much more of the heat of combustion to be transferred to the cooling jacket is most significant at low power output, and up at medium loads this is only somewhat of a problem. The other problem with the radically large in-head pre-combustion chambers is also only somewhat of a problem at medium loads. The small opening between the pre-combustion chamber and the main combustion chamber becomes a large problem under heavier loads up towards to high end of the operating speed range of the engine. The burning mixture of fuel and air being forced out through the small opening assures good mixing of the air and fuel, and this is in fact the only real advantage of the IDI engines. The problem though is that this small opening creates a flow restriction that interferes with the efficiency of the engine somewhat even under medium loads and at medium engine speeds. Under heavier loads up towards the high end of the operating engine speed range the burning mixture of fuel and air has less time to get through the restrictive opening, and frictional losses go up dramatically.
Up near the maximum engine speed under a heavy load there is also a very severe problem with the flow restriction preventing the engine from running well with as late of an injection end timing as would otherwise be possible. At high engine speed the flow restriction causes a large delay between the injection of the fuel and it's entering the main combustion chamber where the last of the un-reacted oxygen is present. What this means is that IDI engines end up having a narrower range of speeds and loads overwhich they run well. In order to get an IDI engine to run anywhere decently at high power output up at the top end of the engine speed range without excessive visible smoke formation the end of injection must be significantly earlier than would be the case with a direct injection engine. This earlier end of injection timing under a maximum load means that either the injection flow rate has to be higher or the injection start timing must be earlier. For an inline injected engine this means much worse operation at reduced speeds and loads. What this all adds up to is the fact that IDI engines running under ideal medium to medium heavy loads at just the right engine speed typically use 20 to 30% more fuel than direct injection engines. Under heavier or lighter loads or at higher or lower engine speeds the difference is much more dramatic, and in practice IDI engines often seem to require nearly twice as much fuel as direct injection engines in the same application.
Since most small marine diesel engines are of the indirect injection type they use way to much fuel however that fuel is injected, but at reduced power output the traditional Bosch type inline injection pumps do a particularly poor job and fuel consumption tends to be unbelievably high. In the larger sizes common rail engines are already available that do far better than most traditional diesel engines and there is no reason that common rail engines cannot also exceed the performance of the best rotary distributor injected engines of the 1950's. The fact that no small diesel engines are currently available is somewhat disturbing, but the only thing that can be done for small sailboat voyaging is to make the best of whatever can be obtained. Unfortunately if all that can be obtained is an indirect injection engine with a Bosch type inline pump then making the best of it probably means motoring at nearly hull speed and burning quite a large amount of fuel. A good reason to have a really tall rig and know how to use it.
Back to Index to Technical Articles