Powertrain system modeling is at the heart of all projects in PCRL, whether they are for simulation, diagnostics or control. However, several research projects within the lab have focused principally on developing appropriate methodologies, structures, and software for dynamic system simulation. These dynamic models have been used for vehicle simulators, as design and feasibility tools, to understand how sub-systems affect each other, how energy is utilized within the system, and to understand transient performance characteristics.
Members of PCRL have developed dynamic engine models for the Iowa Driving Simulator, a $12M motion platform vehicle simulator that is being used for soldier-in-the-loop studies. PCRL has also developed several dynamic SI and CI engine models for the National Advanced Driving Simulator (NADS), a $32M vehicle simulator funded by NHTSA and other federal agencies.
PCRL has also developed a number of powertrain system dynamic models as part of the US Army's Automotive Research Center (ARC). In this 3-year, $7.5M initiative (with the University of Michigan, University of Iowa, Wayne State University and Howard University) PCRL has developed the modular powertrain system models for the M916 Freightliner semi-tractor (2 DDC Series 60 modular diesel engine models including a full cycle simulation, Allison HT-740 automatic transmission with full dynamic powertrain), the HMMWV (DDC 6.5L IDI diesel engine model, Hydra-matic 4L80-E automatic transmission, Zexel-Gleason differentials and 2 dynamic powertrain models), the M1A1 Abrams main battle tank (2 dynamic models of the AGT-1500 gas turbine engine, and full dynamic model of the X1100-3B transmission and powertrain module), and the M2A2 Bradley Fighting Vehicle (600hp Cummins VT903T engine and full hydro-mechanical steering automatic transmission).
Many other dynamic models of both gasoline and diesel engines have been developed by PCRL . These include the Cummins L10, M11, and N14 6-cylinder CI engines, the Ford 4.6L V-8 SI engine and the 3.0L V-6 SI engine, the Buick 3800 V-6 SI engine, the DDC Series 60 DI and 6.2L IDI CI engines, and the International 4.5L V-6 DI CI engine. These dynamic modular engine models are used throughout the USA, as well as in Japan, India, Taiwan, Colombia, Korea, Sweden, Italy, and this program continues to be quite popular. A partial list of customers includes: Daihatsu Motor Co., Delphi Automotive Systems, Denso Co., EDS, Ferrari S.p.A., Fiat Auto R&D USA, Hino Motors, Hitachi America, Jaguar Cars, Komatsu Ltd., Lucerne Institute of Technology, Mitsubishi Electric, Ricardo Consulting Engineers, Samsung Electronics, Seoul National University, Southwest Research Institute, Subaru Research and Design, Synchro-Start Products Inc., Toyota and Woodward Governor Co.
The high-bandwidth, transient dynamometer systems research in PCRL is focused on developing specialized test and development tools for both single and multi-cylinder engines to be used for transient engine research and development. In their final form these systems are a virtual load devices, in which all the dynamic loads that an engine would experience in a vehicle are recreated with the dynamometer system. The multi-cylinder engine test systems are virtual powertrains, and the sincle-cylinder engine test system is a virtual engine. Developing a dynamometer system around the hydrostatic concept has several major advantages. Hydrostatic systems have a very high power density, so the rotational inertia of the hydrostatic system is approximately an order of magnitude lower than comparably powered electric systems. This has several advantages from decreasing engine-dynamometer system resonances to dramatically increasing system slew rates and rotational acceleration. Hydrostatic systems also do not require any modulating high electrical current components near the test engine. All of the high-current components are in the hydraulic power supply, which can be remotely located. This has the advantage of significantly decreasing eletrical noise on sentitive low-current transducers and circuits used to measure engine variables.
Several new inventions have been added to the single-cylinder engine test system in order to replicate the dynamics and transients of the multi-cylinder engine and expand the single-cylinder engine test system's capabilities. These include the dynamic intake air simulator, as well as the transient heat transfer simulator. The intake air simulator reproduces the transient plenum pressure by pulling out air from the plenum at a high bandwidth that would normally go to the other cylinders in a multi-cylinder engine. The plenum pressure is now the same as the multi, and the wave dynamics in the runner are identical to the multi, thus giving the same flow characteristics into the cylinder. The transient heat transfer simulator is a device that consists of a cylinder block with several individual coolant passages, and the temperature and flow in each passage is independently controlled by means of a mixing manifold, flow measurement, and closed loop control. When all of these systems are combined, the single-cylinder engine "thinks" it is in a multi-cylinder engine. Also, all of these devices can simulate very rapid transients. For example the transient dynamometer is limited to 10,000 rpm/sec slew rate, and the pneumatic valves controlling the plenum pressure have an opening time on the order of 0.001 sec to replicate gas flow to the other cylinders. Several patents have either been issued or are pending in the U.S. Patent and Trademark Office on these inventions.
A major technical contribution from PCRL has been the development of nonlinear cylinder pressure and heat release observers that can be used to estimated these parameters on a cylinder-by-cylinder basis from measurements of engine crankshaft velocity. This represents a synthesis of engine cycle simulations (or cylinder-by-cylinder engine models) and nonlinear sliding mode observers. These observers allow the estimation of heat release at all engine speeds, because of the accurate handling of internal inertia effects in the engine.
These observers have three main elements that allow this estimation to occur; a feedforward structure that is essentially an inverse of the nonlinear combustion and rotational engine dynamics, a feedback structure that corrects the estimation based on physical measurements and provides robustness, and the error state dynamics which govern the observer's convergence rate. There are other special structures that allow the observer to be turned off and on for individual cylinders, and to vary the feedback gain near top dead center to address observability issues. Researchers have developed several forms of the heat release and cylinder pressure observers, and the reduced form of cylinder torque observer. Validation of these algorithms using experimental data indicates that these structures exhibit reasonable robustness to modeling errors and variations in heat release rates at various operating conditions.
An observer structure has been developed that is based on a multi-volume model of an engine intake manifold, and estimates air flow rates into each individual cylinder. The goal of this project is to more accurately estimate individual cylinder air charge in order to more closely regulate the bulk air/fuel ratio in each cylinder. This is an important parameter because the conversion efficiency of the 3-way catalyst system may be closely linked to air/fuel ratio or air/fuel ratio time trajectories, and this conversion efficiency has a very strong influence on regulated vehicle tailpipe exhaust emissions.
The dynamic model splits the manifold volume into the plenum and each of the runners. The flow across each of the boundaries is estimated by orifice equations or volumetric efficiency relationships. Additional physical information is needed to obtain these estimates, and intake runner pressure sensors are used to help estimate runner flow rates. The number of additional pressure sensors needed to accomplish these tasks are fewer than half the number of runners, because not all of the runners experience flow at the same time and some physical multiplexing can take place.
There are still some challenges to be overcome in developing these air charge observers because of the simple flow relations being used. Inertial effects need to be included in the models to get reasonable estimates at high engine speed, because of the air pulsing and resonator effects, and volumetric efficiency estimates for transient operation must be evaluated. Other approaches to the air charge observers are being explored, and this can be a useful tool in controlling transient air/fuel excursions.
Diagnostics, or the estimation of input forces on mechanisms with varying inertia, can be quite difficult because of the internal inertial effects that can be very large, cause false detection alarms, or can completely mask internal faults. This is the case with engine diagnostics applied to engines with relatively few cylinders (four or fewer cylinders on a 4-stroke engine, or 2 or fewer cylinders on a 2-stroke engine). These engines can exhibit very large internal inertial effects at high nominal engine speeds, as a result of their typical slider-crank geometry, that can cause large accelerations.
A methodology and implementation has been invented and developed in PCRL that removes the influences of these internal inertial effects from engine speed measurements. This methodology can be implemented in a very cost-effective manner, does not require expensive or special speed transducers, and exhibits excellent robustness charateristics. More importantly, the secondary implications of this approach may more than offset the costs of the algorithm. The methodology is a transformation that makes the engine appear as a linear system in the "synthetic" engine speed signal, and therefore the speed dependency is completely eliminated. Engine control and diagnostics algorithms that use these "synthetic" signals no longer need to have a speed-dependent structure, the same diagnostics strategy and code can be used for all engine speeds, thus dramatically simplifying the structure and length of these algorithms. There is then a tertiary benefit, because of this simplified structure, of lower costs in modifying the control and diagnostic code when such changes become necessary. With the size of control and diagnostic code growing exponentially in time, these benefits can be substantial.
This methodology has application to most medium and small automobile engines, outboard marine engines, lawn care and snow removal equipment, most small to medium size maintenance, utility and construction equipment, and most utility tractors and internal combustion powered tools. It can be used equally effectively on gasoline or diesel engines, because it addresses issues of rotational dynamics. This is an important development because of impending stringent emission regulations on this segment of the engine industry. Without an efficient and cost-effective methodology for addressing these emission regulations, the challenge of meeting these new guidelines could prove financially prohibitive. This invention has been patented with the U.S.P.T.O.
A vehicle's orientation and dynamics are principally controlled by the forces that are generated at the tire-road interface. These forces are modulated by the driver through either steering the wheels, applying energy to the wheels (acceleration), or by braking. If a control system could be devised that would allow independent wheel control over steering and torque (+ and -), what inprovements in vehicle control would result? This is the subject of a research project that includes the development of simple 3-D vehicle dynamics models, as well as multivariable nonlinear control strategies to achieve the desired vehicle orientation through the modulation of individual wheel torque and steering.
Some of the results from this study are included in the PCRL publications. Today several corporations are producing automotive systems that can measure vehicle orientation through yaw-rate sensors, and modify torque to the wheels through extensions to the ABS (anti-lock brake systems) and TCS (traction control systems) currently offered on more expensive vehicles. Some 4-wheel steer systems have been offered commercially, but their use is not widespread because of the costs and need for extensive hardware modifications. Independent wheel steer and torque would feasible on vehicles with individual wheel motors (electric or hydraulic vehicles), but the cost-benefit tradeoff may limit their application.
Brake systems in current production automobiles are the result of a long evolutionary process beginning with the first practical hydraulic brake patent in 1917. While the basic design has many advantages, recent modifications and additions to this system for anti-lock braking and traction control considerably increase the complexity and cost of manufacture. These systems represent a mechanically actuated, pneumatically amplified, electrically modulated, hydraulically transmitted, cable back-up system on most modern vehicles. A fresh look at the braking system design may be justified at this time, especially with respect to electric vehicle applications.
This research project was a collaboration with Chrysler Corporation, Bendix Corporation, and faculty and students from the Mechanical Engineering and Electrical & Computer Engineering Departments. The tasks were to study the design of electric brake systems, to analyze the robustness and fault tolerance of these designs, and to search for new ideas and solutions to this problem. Both Failure modes and effects analysis (FMEA) and Fault Tree Analysis (FTA) were used in this study, and many recommendations were make to the sponsors. The electric brake systems in this project were built by Bendix and implemented on a Chrysler electric van for demonstration purposes.
This project was originally started at MIT, was funded by General Motors Research Laboratories - Power Systems Research Department, and carry-over studies have continued on the basic topic in PCRL. The basis for this research is that in the early and mid 1980's most engines and transmissions were controlled independently, and most transmission control was realized through a complex maze of hydraulic plumbing and actuators. The goals of this research are to analyze the advantages in controlling the engine and transmission in a coordinated manner, and to examine the benefits that could be realized in electronic control of the transmission and clutch-to-clutch shifts. Dynamic models of the components were developed by a research team, and several control algorithms were examined and implemented.
A look at current powertrain controllers and control strategy reveals that most engines and transmission have some level of coordinated control strategy. This may be as simple as decreasing engine torque briefly during the speed or inertia phase of the shift for improving shift quality, to a complete transmission and engine drive-by-wire system (e.g., BMW 750il) with powertrain controllers continuously calculating the required system inputs. Clutch-to-clutch shifts have also become more common in production vehicles, with one of the first applications being the Chrysler mini-van. Most modern transmissions are electronically controlled in order to decrease the high manufacturing costs associated with the hydraulic valve body, spools and dampers on previous vehicles.
Copyright © 1989-2016,Powertrain Control Research Laboratory. All rights reserved.