An image acquisition-and-processing camera system was developed for in-cylinder diagnostics of a single-cylinder heavy duty diesel engine. The engine was equipped with an electronically-controlled common-rail fuel injection system that allowed both single and split (multiple) injections to be studied. The imaging system uses an endoscope to acquire luminous flame images from the combustion chamber and ensures minimum modification to the engine geometry. The system also includes an optical linkage, an image intensifier, a CID camera, a frame grabber, control circuitry and a computer. Experiments include both single and split injection cases at 90 MPa and 45 MPa injection pressures at 3/4 load and 1600 rev/min with simulated turbocharging. For the single injection at high injection pressure (90 MPa) the results show that the first luminous emissions from the ignition zone occur very close to the injector exit followed by rapid luminous flame spreading. For split injection cases, as the amount of fuel in the first injection decreases, ignition occurs further downstream of the nozzle. When only a small amount of fuel is injected, ignition does not occur until the fuel reaches the piston crown. The second fuel injection pulse is injected into a high temperature environment containing the combustion products of the first injection and flame luminosity appears close to the injector exit with almost no detectable delay. Experimental results are compared to numerical computations using a version of KIVA-II code with improved spray, ignition, combustion and emissions models. Good levels of agreement were obtained with measured cylinder pressures, heat release rates, ignition timings and locations and flame shapes. However, the need for some improvement in the vaporization and mixing models is indicated. The present imaging and modeling results are also used to explore the reasons for the significant soot and NOx emissions reductions with multiple/split injections. In this case, a late second injection prevents the high temperatures that promote NOx formation. The combustion products of the first injected fuel set up a sufficiently high temperate for the rapid ignition of the second injection. The second injection may quench NOx being formed from the combustion of the first injection and also entrain air into the first combustion region. This helps in the oxidation of soot.DUE TO ENVIRONMENTAL concerns, emissions standards for engines are becoming more stringent. Thus, engine manufacturers are faced with the need to further develop current technology and/or devise new technologies for emissions control while maintaining engine efficiency. However, the development of new technologies can be costly and time consuming unless adequate tools are at hand. A predictive numerical model could serve as such a tool to help the engine designer with performance and emissions predictions.The objective of this study is to develop a combustion diagnostic system which can serve to validate a predictive combustion model for diesel engines. An endoscope based imaging system is used to visualize combustion in a single-cylinder version of a production diesel engine under different operating conditions. The optical access is designed to be as non-intrusive as possible, so as not to alter the engine flow characteristics. The experimental data includes engine cylinder pressures, ignition timings and locations, and flame development and propagation. The experimental data are also useful to help explain observed trends in emissions reduction through the use of different injection schemes, including both single and split injection schemes, and to assess the accuracy of a predictive model.There are many ways to categorize engine combustion diagnostics. Generally, the light sources for flame and spray images come in two ways: natural flame luminosity and laser-induced scattering *. Images can be taken either with a high-speed camera, providing continuous images, or a digital CCD/CID camera allowing a single-shot for a designated engine cycle. The optical access into the engine allows for further classification of combustion diagnostics methods as direct or light-guided imaging.Direct imaging methods usually require major modifications of the engine geometry, which provide a large field of view, but altered engine flow characteristics. For instance, high-speed cinematography was used to record the natural flame luminosity to study the soot distribution in a diesel engine equipped with a transparent piston by Dec  and Espey and Dec . The natural luminosity images give a picture of luminous flame emission integrated along the line of sight to a depth dependent on the optical density of the plume. Therefore, the information at a specific layer within the spray cannot be obtained. Laser-induced imaging techniques can be employed to provide temporally and spatially resolved measurements. Generally speaking, both the LIF (laser-induced fluorescence) and Exciplex (Excited-State-Complex) fluorescence methods can be used to image the fuel droplets and vapor of a diesel spray , . On the other hand, to image the flame and soot, LII (laser-induced incandescence) and elastic scattering methods are commonly used , , . The LII and elastic scattering images provide the soot distribution in the plane of the laser sheet, which can be positioned at any desired elevation through the plume.Although the above laser-based measurements can provide valuable in-cylinder information, extensive optical windows using transparent materials are required to introduce the laser light into the cylinder. Such a modification may change the flow and combustion characteristics. Less intrusive measurements without major engine modifications can be done through light-guided imaging systems which use fiber-optics or endoscopic devices.Light-guided imaging systems use relatively small windows for imaging to avoid modifying the engine geometry. However, a more complicated optical linkage is needed to transmit the light from the combustion chamber to imaging devices. For example, a randomized fiber bundle together with the two-color method was used to measure the flame temperature and particulate concentration (expressed as the KL number) by Yan and Borman , . An integrated imaging system for flame temperature and soot measurements was developed by Shakal  incorporating a fiber optical bundle with an intensified camera. In the present study an endoscope based system  was used with an intensified CID camera acquisition system, similar to that of Shakal , to study diesel combustion. Luminous flame images were used to better understand the combustion characteristics of various injection schemes and to validate numerical models in KIVA-II code .