Future of Diesel Fuel Injection
During the history of diesel fuel injection, the capabilities and the number of degrees of freedom to optimize diesel combustion have continually improved. Often limits of knowledge, technology or manufacturing capability governed the chain of improvements. The need for technology to meet fuel economy and exhaust emission challenges drove improvements. I’ll provide a brief history and then move to where I believe diesel fuel injection technology might go in the future.
Early systems utilized air-blast principles to effectively atomize and deliver fuel, but were quickly replaced by more efficient and compact “solid injection” systems using only hydraulics. The Bosch “inline pump” is generally recognized as the first system produced in volume by a supplier that could be readily applied to multiple engine designs freeing engine companies from the burden of engineering and manufacturing their own systems. In North America several notable novel systems evolved and were produced in high volume for multiple engines over many years including the Cummins P-T injection system, the General Motors Unit Injectors, and the Roosa-Master distributer pump.
As fuel injection evolved features and functionality were added up and through the mid 1980’s:
- A variety of speed and torque shaping mechanical, hydraulic and electrical governors
- Load based timing control
- Speed based timing control
- Closed vs open nozzles
- Sac volume reduction and VCO nozzles
- Reduced nozzle size including pencil nozzles
In the mid 1980s, a new era began with the introduction of the extensive use of electronics for fuel system control. In 1985 the introduction of the DDEC (Detroit Diesel Electronic Control) set a turning point in the industry with the electronically controlled system providing an extensive range of timing and output control as a function of multiple sensor inputs. The system was quickly followed by a similar systems at other manufactures in both unit injector and distributer pump configurations. These systems also grew in capability with higher injection pressures (over 2000 bar peak pressures by the mid 1990’s in some applications). Pilot injection features were recognized but had limited range capability to starting and idle due to cavitation erosion issues at higher engine speeds. Unit pumps configurations of solenoid controlled unit injector were introduced to the market by Mercedes Benz (produced by Diesel Technology Company) as well as a special version at Mack Truck with Current Controlled Rate shaping which provided a new feature allowing an electronically modulated “boot injection” that could be modulated in duration electronically independent of the state point. VW and Bosch ultimately produced an electronic unit injector for passenger car size engines that had a fixed mechanical pilot injection characteristic built in.
The next major improvement in degrees of freedom came with the Caterpillar/Navistar introduction of HEUI (Hydraulic Electronic Unit Injectors). This was the first production system that provided independent control of injection pressure at all engine loads and speeds, providing particular value to reduce smoke in the lower engine speed range. The system also allowed a wider range in timing independent of the limitations of the fixed cam designs of all previous systems. The system also avoided the risk of live pressure at the nozzle that was recognized with common rail systems.
The introduction of Electronic Common Rail Injection has provide additional degrees of freedom to the diesel fuel system and engine. The concept of common rail injection goes back to early diesel engines in the early 1900’s but was passed over due to practicality of the mechanical controls, reliability, and manufacturing considerations relative to other systems. Robert Huber is credited with the first practical electronic control concept for electrically controlled common rail injectors in the late 1960’s prior to when practical electronic controls were available. I worked with GM built electronic common rail injectors in my early career at Detroit Diesel and extensive single cylinder test work was conducted. This work predated our first demonstration of electronic unit injector in 1979 by several years. At the time we had difficulty building multiple injectors that performed alike and we lacked a practical high pressure pump, using instead pneumatic driven hydraulic pumps to provide pressure. We did however demonstrate stable injection control and ability to independently control a pilot injection. The obstacles initially appeared insurmountable and the standard joke of the time was Common Rail was the fuel system of the future, and always would be.
Denso is credited with the first production common rail system in 1995 with Bosch to follow in 1997. These systems are providing increased degrees of freedom with both independent control of pressure at continually increasing levels as well as multiple injection capabilities for noise and combustion control. Today application of piezoelectric control in place of solenoids provides a even higher levels of precision and rail pressures are heading towards the 3000 bar level in future years.
With the final addition of new degrees of freedom of :
- Variable injection pressure
- Multiple injection capability
- Ultra high pressure
Where might future diesel fuel systems go? There is no doubt that the continued development of higher levels of precision and accuracy will continue as they have in the past. While there has been limited use of combustion feedback through cylinder pressure transducers, this has the potential of being an important enabler of advanced combustion strategies involving premixed combustion (HCCI, RCCI…). But, what are the new frontiers for fuel injection?
Rate shaping has always been a subject of research to improve combustion and or engine noise with a variety of minor implementations over many decades. Nozzle holders with 2 stage lift and the Stanadyne RSN nozzles being typical examples of mechanical systems with limited authority to control initial injection rate.
Around 2001 Mack Truck introduced a variable boot injection system CCRS (Current Controlled Rate Shaping) that perhaps is the length of a boot injection could be electronically controlled and programmed at all engine loads and speeds. Obviously, limitations of the cam profile and varying engine speed constrained the system as well as the calibration needed to control each pump for the desired accuracy.
Nozzles with two sets of spray holes controlled by dual axial concentric needles has for many years been a design of interest (reference slide 10 https://www.energy.gov/sites/prod/files/2014/03/f9/2004_deer_busch.pdf ). Although prototypes have been built and tested with positive results the manufacturability of the design is generally assessed as very difficult or costly. As a result this design has also not seen production use.
With common rail injection, the Delphi direct acting piezo injector offered the best potential for rate shaping with common rail injectors, however the system was unsuccessful in penetrating the market due to technical problems, apparently due to technical problems affecting reliability. Continentals low leakage direct acting piezo design common rail is perhaps the best other recent offering, albeit it also does not appear to have penetrated the market. (I would appreciate contact if there is activity on this I am not aware of).
Multi-Pulse Rate Shaping, sometimes referred to as the “Christmas Tree” approach to rate shaping using multiple injections has been obvious for some time as a means of controlling injection rate and emulating the “rising triangular” rate shape of many unit injection system.
This type of approach may be calibrated using varying degrees of of separation and pulse widths to create the desired injection duration and burn rate to achieve the desired combustion characteristic which may be noise, cylinder pressure, emissions or performance related. Low injection rate results from the upstream throttling of the flow thereby reducing the injection pressure at the spray hole entrance. To effectively use this type of rate shaping, software to control the size, number and separation of the pulses as injection pressure, load and speed are changed.
Obviously to be effective this type of rate shaping needs to be matched with the appropriate nozzle flow and number of spray holes. As discussed in my “Swirl” section, longer duration will likely match best with lower levels of swirl and or fewer spray holes. This technique may best be applied to part load where the injection duration may be shorter than desired.
A possible disadvantage of this type of rate shaping is that there are multiple ends of injection normally related to poor end of injection atomization or “large trailing droplets. An interesting piece of future research would be for somebody to compare this type of rate shaping to a true continual rate shape having the same nozzle geometry, and duration and approximate mean injection rate vs time as roughly illustrated below.
Although such continual rate shapes could potentially be achieved with common rail injectors with direct (piezo) needle control, an obstacle that is present is the high sensitivity to needle lift such a technique would have resulting in injector to injector variation as well as variation with time. Feedback from combustion pressure monitoring might be an effect means of controlling this, however and alternative would be to desensitize the injector design.
Desensitizing the injector can be accomplished by known means of adding a throttling section in the nozzle similar to that used in Stanadyne’s RSN nozzle design.
The throttling section can be placed above or below the lower guide and can be of straight tapered or stepped design and would be designed to be engaged for some fraction of the total needle lift for example 25%. It is particularly noteworthy that this technique can be applied to both piezo and solenoid injectors with ballistic needle control! At lifts above the design value full injection rate would be achieved and below this the the flow would be highly throttled and or throttled in proportion to the shape of throttling section (straight tapered, stepped…).
Ideally for an injector of the non “direct acting type”, this type of a nozzle would employ control piston orifice optimization for slower needle opening, to further reduce the gain and faster needle closing. The faster needle closing is desired to minimizing the throttle effect on the ultimate end of injection needle closure both due to the shortened engagement time with the throttling section and due to the closing needle action as a piston to maintain injection pressure below the throttled region of the nozzle.
More importantly the rate shaping may be enhanced by multi-pulsing the control to essentially “dwell” the needle motion in the throttling section to provide extensions of the rate shaping. If the the throttling section is tapered or stepped, multiple rates may be controlled. Duration of the reduced rate period may be controlled by the length and number of pulses used to effectively dwell of “dither” the needle motion.
This type of rate shaping was identified as “optimal” in a Bosch presentation at the 2004 DEER Conference (refer to slide 10 https://www.energy.gov/sites/prod/files/2014/03/f9/2004_deer_busch.pdf ).
Variable nozzle orifice area can also achieved a similar form of rate shaping for single hole spray nozzles with axial sprays by designing the rate shaping throttle below the needle seat in the form of a pintle nozzle.
This nozzle is discussed in https://riunet.upv.es/handle/10251/82103 in a paper Study of New Prototype Pintle Injectors for Diesel Engine applications and concept illustrations are shown in US Patent Application US20110240770. The interesting aspect of this type of rate shaping is that is done with variable area of the nozzle orifice itself and not upstream throttling thus full rail pressure is available across the orifice at all times. As with the previous discussion of the previous design, many geometric variations of the throttling area of the pintle can be used Several variations of this design was built and tested by my team at General Motors and culminating with the previously mentioned work and paper with the University of Valencia (Ref. Prof Raul Payri) mentioned earlier.
Obviously this concept cannot be utilized with combustion systems designed for multi-hole nozzles. They can however be used in opposed piston engines or large uniflow 2-stroke diesels with multiple injectors around the periphery of the bore.
It is interesting to note that many of the successful diesel opposed piston engines produced employed a single injector per cylinder with a single fixed spray orifice. Obviously the addition of a degree of freedom of rate control could provide potential enhancements to combustion control in such engines. This is particularly relevant with recent interests in opposed piston diesel technololgy (Achates Power, EcoMotors, Superior/Gemini). Whether a single injector is used per cylinder or a pair of injectors, the concept is relevant. If the use of a single variable area nozzle could achieve the performance of a pair of conventional injectors per cylinder, this could provide a significant improvement in cost and serviceability.
The Fairbanks Morse Model 38 8 1/8 engine has used two nozzle holders with spring closed pintle nozzles with mechanical unit pumps. References show the engine has used an outwardly opening pintle nozzle and presumably in recent years a more conventional inwardly opening pintle nozzles. I would assume the recent Achates Power upgrade of this Fairbanks engine uses two conventional multi-hole common rail nozzles similar to published information on their combustion system. It would be interesting to see how how pintle nozzles on high pressure common rail injectors would perform.
The Variable Area Poppet Nozzle is another example of a design with potential for the future with capabilities of variable rate shaping. My first experience with such a design dates to the late 1970’s when as a young engineer and pursued such design for a unit injector with great enthusiasm! My creative dreams were dashed when I found the basics of “Stretch Valve” design (ref Patent US 4,046,322 were already conceived by Philip Scott of the “Super Diesel Tractor Company” in 1921 (reference patent US 1,609,578).
The design was tested on an injection bench with rate tube injection measurement as well as in a spray visualization fixture. The injection rate diagrams confirmed the predicted performance of significantly increasing injection rate as the pressure rose and the flow area increased. However, while the spray visualization fixture showed a 360 degree spray, it was highly eccentric indicating insufficient quality of the prototype parts.
The nozzle was engine tested and in one cylinder of a test engine as well as in a combustion photography engine. The combustion results were exceeding bad with the combustion photography engine revealing a spray not near reaching the bowl periphery. The only positive outcome was the designed showed the potential to have low HC emissions due to absence of both sac and hole volume!
The Variable Area Poppet Nozzle concept has been reconsidered within the context of very high injection pressure common rail injectors. Several factors make this worth considering.
- The high levels of injection pressure can provide significantly more energy and momentum to the spray to improve spray penetration.
- With direct electronic control of the lift variable area functions as well as multi-pulsing techniques can be employed.
- With combustion techniques as HCCI or RCCI, there may be advantages to the uniform 360 degree spray and low penetration of the early injection events.
- The potential for low HC emissions exists with design.
- There are already trends towards higher number of spray orifices with lower levels of swirl. The poppet suggests an end point of the trend in a zero swirl engine
- Small displacement diesels, i.e. less than 0.25 liter per cylinder, might benefit from such a design due to the short spray penetration requirements.
A design of such a nozzle was experimented with late in my career and is a subject of ongoing interest and research. The design explored is described in patent US9261060.
This design was built and tested in both injection rate fixture bench testing and spray visualization tests with high pressure nitrogen. A Continental piezo actuator from a production injector was used for direct actuation both mechanically connected and connected through a hydraulic coupling/amplifier. Relatively symmetric spray patterns were achieved on all but extremely small strokes. By using the piezo actuator stroke and therefore rate shape could be controlled by control of the applied voltage.
One of the features of the design is that the poppet sealing force increases with injection pressure. As a general heuristic, the contact force on the poppet seat needs to be greater than the injection pressure one is trying to seal by a factor related to the precision of the sealing surfaces and their geometric relationship to each other. By making the piston, which pulls the needle against it’s seat, slightly larger than than the poppet seat inner diameter, this can be accomplished.
I would envision an improved version of the injector to incorporate a piezo or magnetostrictive actuator located much nearer to the nozzle piston assembly to avoid some of the deformation and thermal expansion issues that exist. Use of a hydraulic amplifier could both aid in the packaging and performance of a longer thinner actuator as well as provide hydraulic “length adjustment” to compensate for mechanical and thermal length variations.
There is still much work to do to understand the combustion chamber configuration which would be optimum for a poppet injector as well as the pressure requirements and rate shape. It is obvious for conventional diesel combustion that the chamber must provide volume above and or below the spray for the gases in the chamber to move inward as the spray moves outward with the spray itself reversing direction at the bowl periphery. Should the spray be targeted high , low or in the middle???
For advanced diesel combustion systems like HCCI or RCCI, where substantial injection takes place far before TDC, The poppet type nozzle may be of significant advantage due to it’s symmetry and low spray penetration. Such systems also may not have the same need for extremely high injection pressure and as such make the construction and operation of such and injector much easier.