Nevada deploys first US commercial, grid-connected enhanced geothermal system
Toyota wraps up 3-year Strasbourg Prius PHV demo project; average 46% reduction in fuel consumption compared to gasoline ICE vehicles

Delphi advancing Gasoline Direct-Injection Compression-Ignition engine concept; new two-stage supercharger/turbocharger boost system

Simulated fuel economy for 1.8L GDCI Engine and 2.4L GDi baseline engine. GDCI showed a predicted fuel economy improvement of 65%, 28%, and 28% for the FTP City, Highway, and US06 cycles, respectively, for a combined FE improvement of 50%. Source: Sellnau et al. Click to enlarge.

At SAE World Congress next week in Detroit, Delphi Automotive will present two technical papers describing its ongoing progress with the Gasoline Direct-Injection Compression-Ignition (GDCI) engine concept. (Earlier post.)

GDCI is an advanced low-temperature combustion concept that uses compression ignition under lean to near stoichiometric fueling conditions over the complete engine operating range. Previous studies of GDCI have shown good potential for very high efficiency, low NOx, and low PM over the full speed-load range. GDCI achieves low-temperature combustion using multiple-late injection (MLI), intake boost, and cooled EGR.

GDCI combines aspects of diesel and spark-ignited engine technology. The compression ratio is high with multiple-late-injections (MLI)—similar to diesel—but commercial gasoline fuel is used that vaporizes and mixes quickly at low injection pressure, unlike diesel.

The injector is central-mounted, with a symmetrical chamber and piston bowl. The engine is operated unthrottled and diluted with excess air or EGR, depending on load.

Left: GDCI injection strategy depicted on Φ-T diagram with CO concentration. To achieve concurrent low NOx and low PM emissions, combustion must occur in the area bounded by the green box (away from soot and NOx formation regions).
To avoid efficiency-compromising CO emissions, combustion must occur in the region 0 < Φ < 1 with 1300 < T < 2200 K. A primary attribute of this injection strategy is low fuel injection pressure.
Source: Sellnau et al. Click to enlarge.

The work being presented at this year’s World Congress describes the development of a boost system for GDCI and GDCI part-load operation.

Two-stage supercharger/turbocharger system with two liquid charge air coolers (LCCA) to control intake air temperature (IAT) for GDCI. Hoyer et al. Click to enlarge.

Boost system. Like other high-efficiency concepts, GDCI has low exhaust temperatures; unlike some others, however, GDCI relies heavily on fuel injection strategy to ensure proper combustion, the Delphi engineers noted. Since the injection strategies depend on the in-cylinder pressure, temperature, and composition conditions, the injection is coupled to the boost system.

For the study—the goal of which was to demonstrate a practical and efficient boost system for a low-temperature 1.8L, 4-cylinder GDCI engine—the team assumed that the injection strategy will permit proper combustion provided the boost system can deliver the correct air and EGR mixture at the correct pressure and temperature.

They used simulation tools to select the boost architecture, to develop the selected system for full-load and part-load requirements, and to evaluate transient response. They also developed a methodology for simulation-based calibration to determine the best settings for efficient operation and potential for low fuel consumption over the full speed-load map.

They found that a two-stage supercharger/turbocharger system that also has two charge air coolers could supply the needed air charge to meet the GDCI requirements for full-time operation. Other boost system architectures could not meet the GDCI performance targets; turbocharger system performance was degraded by low temperature exhaust, high EGR, and LIVC operation.

The undertook a design effort to package practically the two devices and two coolers into a vehicle. Interactions of the devices affect the performance and efficiency of the system. When the proper combination of supercharger drive ratio and turbine geometry settings are chosen, the simulated results demonstrate that this system can operate efficiently at full load. For part load operation, results show that turbine rack settings and supercharger bypass valve can be used to achieve low BSFC.

Preliminary transient simulations of the system for an aggressive acceleration from idle to full load indicates potential for reasonably fast response, they found. Real transient response will depend on the engine control system.

Overall, they concluded, simulation results indicate the GDCI engine with the proposed two-staged boost system should meet performance objectives with good transient response.

Part-load operation. In the part-load operation study, which involved both a single-cylinder and multi-cylinder engine, Delphi engineers designed a new piston to match with the injection system. Piston design requirements for GDCI are different than for diesel, they noted. The new piston supports a near quiescent chamber with low squish and low or no swirl. This is desirable to maintain stratification created and controlled by the injection process and to avoid overmixing of fuel and air.

Delphi matched the bowl shape and fuel spray for typical GDCI injection timing. This resulted in significantly reduced piston surface area with less propensity for fuel wall wetting. Low-temperature GDCI combustion performs best when injected fuel stays in the hot bulk gases for complete oxidation and minimal heat losses to piston and liner surfaces, the team noted.

With the fuel injection system and the new piston design, the GDCI combustion system demonstrated very good fuel efficiency and emissions in single-cylinder engine testing. At part-load conditions, injection pressure could be lowered significantly, while achieving low fuel consumption, NOx, PM, and combustion noise.

Heat losses and combustion losses were both very low, and contributed to indicated thermal efficiencies of approximately 47%, the Delphi team said. Losses associated with CO and HC emissions were higher than desired and are strong candidates for near term work.

Particulate emissions, measured using a TSI particle size spectrometer, were near levels for filtered room air. The Delphi team suggested that there is good potential to eliminate aftertreatment for PM and NOx species; however, an oxidation catalyst for HC and CO is expected to be needed.

Other findings of the study included:

  • At part-load conditions, boost pressure and EGR could be lowered with improved combustion stability. COV IMEP and variation of CA50 and were both within acceptable ranges. Tests showed that injection parameters could be used to control combustion phasing and other combustion parameters.

  • Good low-speed BMEP was predicted with maximum BMEP of 20 bar at 1800 to 2000 rpm. The system lends itself to aggressive down-sizing, down-speeding, and up-loading for the most efficient operation.

  • Vehicle simulations for various drive cycles predicted very good fuel economy for all city and highway modes. Predicted fuel economy was improved 65%, 28%, and 28% for the FTP City, Highway, and US06 cycles, respectively, for a combined FE improvement of 50%. For the NEDC, EUDC, and WLTC cycles, fuel consumption (L/100km) decreased 37%, 27%, and 33%, respectively.

  • On a multi-cylinder engine, preliminary parametric tests at 2000 rpm-11 bar IMEP showed ISFC and combustion noise were low at 175 g/kWh and 88 dB, respectively, and were comparable to that measured on the single-cylinder engine. NOx and smoke emissions were low and below targets, but HC emissions were above expectations. More work is needed to characterize the engine over the operating map, conduct cold starts and transient tests, and optimize systems for minimum fuel consumption at target emissions levels, the team concluded.


  • Hoyer, K., Sellnau, M., Sinnamon, J. and Husted, H. (2013) Boost System Development for Gasoline Direct- Injection Compression-Ignition (GDCI). SAE Int. J. Engines 6(2) doi: 10.4271/2013-01-0928

  • Sellnau, M., Sinnamon, J., Hoyer, K., Kim, J., Cavotta, M., and Husted, H. (2013) Part-Load Operation of Gasoline Direct-Injection Compression Ignition (GDCI) Engine. SAE 2013-01-0272 doi: 10.4271/2013-01-0272



I always thought that was the way to go with gazoline engine, I know it is a difficult concept to implement because it requires tight control of injection and all parameters, but it carries big rewards in term of efficiency...

I think it is the next big thing in gazoline ICE


With such great performance specs, everyone should be asking when it can hit showrooms.  We need it, badly.

Roger Pham

Very impressive results. No doubt will meet Tier 2 bin 5 NOx standard, or even bin 3 with more tweaking and EGR. However, I wonder how would this new combustion strategy meet the proposed Tier 3 or LEV 3 emission standard for NOx + Smog? Lean combustion is more efficient than stoichiometric, however, NOx clean up is the Achille's heel of lean combustion due to the cost of LNT and the cost and complexity of SCR used in Diesel.



Mazda have diesel that pass without NOx clean up using reduced CR, so I guess it should work here too

Roger Pham

I know, GCCI will pass Tier2 Bin5, no doubt. The question is how about Tier 3 for LDV?


Wonderful to see the efflorescence of a dying technology.What will those engineers come up with next?This one is crying out for TIGERS.An electric supercharger.


Every one of these 'advanced' ICE components - IC controller families, turbos, fuel injection,.. - was available decades ago, yet CAFE mpg remained the historic 80 year average ~20 mpg.

EVs(~100 mpge) are mass produced less than three years and ICE CAFE improves by over 20% to >24+ mpg already.

The ignorance and arrogance of the anti-EV/oil terrorists is obvious and criminal.


Had Hybrids and EVs not been crushed-canned in the late 90s, ICEVs and EV batteries evolution would have been 10+ years earlier. CAFE 60 mpg ICEVs would be a reality and 100+ mpge Electrified vehicles would already have 30+% of the market.

Gasoline consumption would be down 30+%.

This 10+ years delay was very beneficial to the Oil industry but almost killed the Big 3 or should it be the Big 2 because Ford is catching up fast.

Every one of these 'advanced' ICE components - IC controller families, turbos, fuel injection,.. - was available decades ago

Every component of an Indy car (ICE, transmission, etc.) was available when the Model T was built, too.

Today's advances in efficiency rely on advances in sensors and actuators as well as the computing power to make use of them.  20 or even 10 years ago, sufficiently powerful computers were large, expensive and fragile affairs that could not be relied upon to run critical vehicle systems.  If you go back to the 1980's and look at the processing speeds of engine ECUs (e.g. the time required to do a 16-bit integer addition), you would first shake your head in disbelief and then nod in agreement with the people telling you it just could not be done then.

I know.  There is still code out there that I've touched, making those early-generation computerized drivetrains go.  Some of it is just a little bit faster and more capable because of my careful implementation.

Had Hybrids and EVs not been crushed-canned in the late 90s, ICEVs and EV batteries evolution would have been 10+ years earlier.

Had PNGV not been canned by Bush-43, we'd have had PHEVs 5-7 years earlier.


"Every component of an Indy car (ICE, transmission, etc.) was available when the Model T was built, too."

Sorry, no model T turbos, fuel injection,..

The Pentium processor was on sale in 1993, advanced sensors got man to the moon in 1969.

The 1991 Indy car winning speed was 176 mph.

The 2012 Indy winning speed was 167 mph.(wiki)

As I said, "Every one of these 'advanced' ICE components - IC controller families, turbos, fuel injection,.. - was available decades ago, yet CAFE mpg remained the historic 80 year average ~20 mpg.

EVs(~100 mpge) are mass produced less than three years and ICE CAFE improves by over 20% to >24+ mpg already.

The ignorance and arrogance of the anti-EV/oil terrorists is obvious and criminal."


Sorry, I should have said NASCAR cars.  No turbos, no fuel injection.

The Pentium was far too expensive and required far too much bulky, fragile and heat-intolerant external hardware to run an engine.  I know; I was paid to convert Pentium code to run on something that could tolerate the environment of a transmission.

The sensors used in Apollo had nothing on an early-90's MAP sensor.  The intelligence that is built into some 2010+ sensors makes the latter look like a Down's victim compared to Einstein.

We are indeed seeing the last gasps of the internal combustion engine.  That they are the last, does not make them the less remarkable for the computer power that makes them possible for the first (and last) time.

Roger Pham


With continual refinement to the ICE like those made by Delphi in this article, perhaps the ICE will hang around for quite a long time. A highly efficient ICE coupled with a HEV drive train can use synthetic hydrocarbon fuel made from RE and become sustainable practice. May not be as efficient as a BEV or FCV on a source to wheel calculation, but many consumers may prefer ICE-HEV over BEV or FCV and may be willing to pay more for fuel cost.


I expect the ICE to remain in specialty applications for decades to come, if only because the cost of retrofitting small-volume legacy systems won't be worth it.  But all it will take is a handful of robust SOFCs or other products to take most of the new applications away from ICEs, even those which remain reliant on liquid fuels.


Not sure if I understand enough, but since the pollution issues were discussed under the partial load section, can I safely assume that pollution is not an issue at full load? If so then it is a good candidate as a generator in a series hybrid car as is.


My choices on my last car saved me about 1800 gallons of fuel, which more than paid for the extra powertrain cost.

I'm expecting the same with this one.

Roger Pham

At full load, expect higher NOx due to higher combustion temp and less EGR, depending on how "full" you want your max load to be, but lower CO and HC due to better combustion. Some sort of NOx reduction will be needed to meet current and future NOx emission regulation if the engine is to be used at high load for significant amount of time.

However, as extended-range ICE for a large-battery-pack PHEV, the mfg perhaps can negotiate with the EPA as to how often the ICE will be used, and hence how much overall NOx will be released per mile overall.

Serial hybrid EV is not an efficient HEV arrangement in comparison to a serial-parallel hybrid arrangement in which the serial hybrid mode is used only at low speed or high speed, while cruise from 30-80 mph can use the parallel mode for increase efficiency. HEV with a small motor and small battery pack uses strictly parallel mode with a geared transmission for maximum utilization of the small motor's torque at low speeds.


GDCI and other more efficient fuel consumption technologies for the ICE will come.

Despite the Conspiracy theorists and assorted Luddite posters that are always here to post their screeds, the automobile is evolving as engineers work hard to improve it, and the auto is becomiing non-polluting.

Perfection is unlikely, but for all practical purposes the auto is/will improve to the point that the Air is not polluted; and the concerns become purely academic. Indeed the Air across the USA has cleared using vehicles compying with T2B5.

Roger as a conscientious engineer, is quite right to wonder about meeting Lev III SULEV II, aka T2B2 levels, or in layman's terms EV levels of emissions.

But it wasn't the know-nothing lawyer-politicians who suggested reducing emissions to that level. It was the automakers who knowing it can be done, proposed it in lieu of more onerous regulations dreamed up by lawyer-politicians doing engineeering without a license.

But as always the demgogues can and will double-cross and demand that, and still maintain their wish ideas for non-reality. These know-very-littles are always willing to set requirements without a thought on reality.

Its too bad that the lawyesr succeed in exempting themselves from malpractice torts, while imposing their 33% fee-taxes on any errors committed by everyone else. What is sauce for the goose should also be sauce for the legal gander.

The comments to this entry are closed.