With stricter emission legislations and customer demands on low fuel consumption, good control strategies are necessary. This may involve control of variables that are hard, or even impossible, to measure with real physical sensors. By applying estimators or observers, these variables can be made available. The quality of a real sensor is determined by e.g. accuracy, drift and aging, but assessing the quality of an estimator is a more subtle task. An estimator is the result of a design work and hence, connected to factors like application, model, control error and robustness.

The air mass-flow in a diesel engine is a very important quantity that has a direct impact on many control and diagnosis functions. The quality of the air mass-flow sensor in a diesel engine is analyzed with respect to day-to-day variations, aging, and differences in engine configurations. The investigation highlights the necessity of continuous monitoring and adaption of the air mass-flow. One way to do this is to use an estimator. Nine estimators are designed for estimation of the air mass-flow with the aim of assessing different quality measures. In the study of the estimators and quality measures it is evident that model accuracy is important and that special care has to be taken, regarding what quality measure to use, when the estimator performance is evaluated.

Maps or look-up tables are frequently used in engine control systems, and can be of dimension one or higher. Their use is often to describe stationary phenomena such as sensor characteristics or engine performance parameters like volumetric efficiency. Aging can slowly change the behavior, which can be manifested as a bias, and it can be necessary to adapt the maps. Methods for bias compensation and on-line map adaptation using extended Kalman filters are investigated and discussed. Key properties of the approach are ways of handling component aging, varying measurement quality, as well as operating point dependent model quality. Handling covariance growth on locally unobservable modes, which is an inherent property of the map adaptation problem, is also important and this is solved for the Kalman filter. The method is applicable to off-line calibration ofmaps where the only requirement of the data is that the entire operating region of the system is covered, i.e. no special calibration cycles are required. Two truck engine applications are evaluated, one where a 1-D air mass-ffow sensoradaptation map is estimated, and one where a 2-D volumetric efficiency map is adapted, both during a European transient cycle. An evaluation on experimental data shows that the method estimates a map, describing the sensor error, on a measurement sequence not specially designed for adaptation.

A systematic design method for reducing bias in observers is developed. The method utilizes an observable default model of the system together with measurement data from the real system and estimates a model augmentation. The augmented model is then used to design an observer which reduces the estimation bias compared to an observer based on the default model. Three main results are a characterization of possible augmentations from observability perspectives, a parameterization of the augmentations from the method, and a robustness analysis of the proposed augmentation estimation method. The method is applied to a truck engine where the resulting augmented observer reduces the estimation bias by 50% in a European transient cycle.

A method for bias compensation in model based estimation utilizing model augmentation is developed. Based on a default model, that suffers from stationary errors, and measurements from the system a low order augmentation is estimated. The method handles models described by differential algebraic equations and the main contributions are necessary and sufficient conditions for the preservation of the observability properties of the default model during the augmentation.

A characterization of possible augmentations found through the estimation, showing the benefits of adding extra sensors during the design, is included. This enables reduction of estimation errors also in states not used for feedback, which is not possible with for example PI-observers. Beside the estimated augmentation the method handles user provided augmentations, found through e.g. physical knowledge of the system.

The method is evaluated on a nonlinear engine model where its ability to incorporate information from additional sensors during the augmentation estimationis clearly illustrated. By applying the method the mean relative estimation error for the exhaust manifold pressure is reduced by 55 %.

A method for bias compensation in model based estimation utilizing model augmentation is developed. Based on a default model, that suffers from stationary errors, and measurements from the system a low order augmentation is estimated. The method handles models described by differential algebraic equations and the main contributions are necessary and sufficient conditions for the preservation of the observability properties of the default model during the augmentation.

A method for bias compensation in model based estimation utilizing model augmentation is developed. Based on a default model, that suffers from stationary errors, and measurements from the system a low order augmentation is estimated. The method handles models described by differential algebraic equations and the main contributions are necessary and sufficient conditions for the preservation of the observability properties of the default model during the augmentation. A characterization of possible augmentations found through the estimation, showing the benefits of adding extra sensors during the design, is included. This enables reduction of estimation errors also in states not used for feedback, which is not possible with for example PI-observers. Beside the estimated augmentation the method handles user provided augmentations, found through e.g. physical knowledge of the system. The method is evaluated on a nonlinear engine model where its ability to incorporate information from additional sensors during the augmentation estimation is clearly illustrated. By applying the method the mean relative estimation error for the exhaust manifold pressure is reduced by 55%.

When estimating states in engine control systems there are limitations on the computational capabilities.This becomes especially apparent when designingobservers for stiff systems since the implementation requires small step lengths. One way to reduce the computational burden, is to reduce the model stiffness by approximating the fast dynamics with instantaneous relations, transformingan ODE model into a DAE model.

Performance and sample frequency limitations for extended Kalman filters based on both the original ODE model and the reduced DAE model for a diesel engine is analyzed and compared. The effect of using backward Euler instead of forward Euler when discretizing the continuous time model is analyzed.

The ideas are evaluated using measurement data from a diesel engine.The engine is equipped with throttle, EGR, and VGT and the stiff model dynamics arise as a consequence of the throttle between two control volumes in the air intake system. It is shown that even though the ODE, for each time-update, is less computationally demanding than the resulting DAE, an EKF based on the DAE model achieves better estimation performance than one based on the ODE with less computational effort. The main gain with the DAE based EKF is that it allows increased step lengths without degrading the estimation performance compared to the ODE based EKF.

A method for bias compensation and online map adaptation using extended Kalman filters isdeveloped. Key properties of the approach include the methods of handling component aging, varyingmeasurement quality including operating-point-dependent reliability and occasional outliers, andoperating-point-dependent model quality. Theoretical results about local and global observability,specifically adapted to the map adaptation problem, are proven. In addition, a method is presented tohandle covariance growth of locally unobservable modes, which is inherent in the map adaptationproblem. The approach is also applicable to the offline calibration of maps, in which case the onlyrequirement of the data is that the entire operating region of the system is covered, i.e., no specialcalibration cycles are required. The approach is applied to a truck engine in which an air mass-flowsensor adaptation map is estimated during a European transient cycle. It is demonstrated that themethod manages to find a map describing the sensor error in the presence of model errors on ameasurement sequence not specifically designed for adaptation. It is also demonstrated that themethod integrates well with traditional engineering tools, allowing prior knowledge about specificmodel errors to be incorporated and handled.

A method for online map adaptation is developed. The method utilizes the EKF as a parameter estimator and handles parameter aging, operating point dependent model and measurement quality. Map adaptation, by construction, gives marginally stable models with locally unobservable modes, that are handled. The method is also suitable for offline calibration of maps where the only requirement of the data is that the entire operating region of the system is covered. The method is applied to a truck engine where an air mass-flow sensor adaptation map is estimated based on data from a diesel engine during an ETC. It is shown that an adaptation map can be found in a measurement sequence not specially designed for adaptation.

A model-based misfire detection algorithm is proposed. The algorithm is able to detect misfires and identify the failing cylinder during different conditions, such as cylinder-to-cylinder variations, cold starts, and different engine behavior in different operating points. Also, a method is proposed for automatic tuning of the algorithm based on training data. The misfire detection algorithm is evaluated using data from several vehicles on the road and the results show that a low misclassification rate is achieved even during difficult conditions. (C) 2014 Elsevier Ltd. All rights reserved.

Four methods for compression ratio estimation based on cylinder pressure traces are developed and evaluated for simulated and experimental cycles. Three methods rely upon a model of polytropic compression for the cylinder pressure. It is shown that they give good estimates with a small bias at low compression ratios. A variable projection algorithm with a logarithmic norm of the cylinder pressure yields the smallest confidence intervals and shortest computational time for these three methods. This method is recommended when computational time is an important issue. The polytropic pressure model lacks information about heat transfer and therefore the estimation bias increases with compression ratio. The fourth method includes heat transfer, crevice effects, and a commonly used heat release model for firing cycles. This method estimates the compression ratio more accurately in terms of bias and variance. The method is more computationally demanding and thus recommended when estimation accuracy is the most important property. In order to estimate the compression ratio as accurately as possible, motored cycles with high initial pressure should be used.

Three methods for estimating the compression from measured cylinder pressure traces are described and evaluated for both motored and fired cycles against simulated and measured cylinder pressure. The first two rely upon a model of polytropic compression, and it is shown that they give a good estimate of the compression ratio for simulated cycles for low compression ratios. For high compression ratios, these simple models lack the information about heat transfer. The third method includes a standard heat transfer and crevice effect model, together with a heat release model and is able to estimate the compression ratio more accurately.

Four methods for compression ratio estimation based on cylinder pressure traces are developed and evaluated for both simulated and experimental cycles. The first three methods rely upon a model of polytropic compression for the cylinder pressure. It is shown that they give a good estimate of the compression ratio at low compression ratios, although the estimates are biased. A method based on a variable projection algorithm with a logarithmic norm of the cylinder pressure yields the smallest confidence intervals and shortest computational time for these three methods. This method is recommended when computational time is an important issue. The polytropic pressure model lacks information about heat transfer and therefore the estimation bias increases with the compression ratio. The fourth method includes heat transfer, crevice effects, and a commonly used heat release model for firing cycles. This method is able to estimate the compression ratio more accurately in terms of bias and variance. The method is more computationally demanding and is therefore recommended when estimation accuracy is the most important property. © 2005 Elsevier Ltd. All rights reserved.

The objective is to investigate models of the specific heat ratio for the single-zone heat release model, and find a model accurate enough to introduce a modeling error less than or in the order of the cylinder pressure measurement noise, while keeping the computational complexity at a minimum. Based on assumptions of frozen mixture for the unburned mixture and chemical equilibrium for the burned mixture, the specific heat ratio is calculated using a full equilibrium program for an unburned and a burned air-fuel mixture, and compared to already existing and newly proposed approximative models of γ.

A two-zone mean temperature model, Matekunas pressure ratio management and the Vibe function are used to parameterize the mass fraction burned. The mass fraction burned is used to interpolate the specific heats for the unburned and burned mixture, and then form the specific heat ratio, which renders a small enough modeling error in γ. The specific heats for the unburned mixture is captured within 0.2 % by a linear function, and the specific heats for the burned mixture is captured within 1 % by a higher-order polynomial for the major operating range of a spark ignited (SI) engine.

Model based diagnosis and supervision of industrial gas turbines are studied. Monitoring of an industrial gas turbine is important as it gives valuable information for the customer about service performance and process health. The overall objective of the paper is to develop a systematic procedure for modelling and design of a model based diagnosis system, where each step in the process can be automated and implemented using available software tools. A new Modelica gas media library is developed, resulting in a significant model size reduction compared to if standard Modelica components are used. A systematic method is developed that, based on the diagnosis model, extracts relevant parts of the model and transforms it into a form suitable for standard observer design techniques. This method involves techniques from simulation of DAE models and a model reduction step. The size of the final diagnosis model is 20% of the original model size. Combining the modeling results with fault isolation techniques, simultaneous isolation of sensor faults and fault tolerant health parameter estimation is achieved.

Supervision of the performance of an industrial gas turbine is important since it gives valuable information of the process health and makes efficient determination of compressor wash intervals possible. Slowly varying sensor faults can easily be misinterpreted as performance degradations and result in an unnecessary compressor wash. Here, a diagnostic algorithm is carefully combined with non-linear state observers to achieve fault tolerant performance estimation. The proposed approach is evaluated in an experimental case study with six months of measurement data from a gas turbine site. The investigation shows that faults in all gas path instrumentation sensors are detectable and isolable. A key result of the case study is the ability to detect and isolate a slowly varying sensor fault in the discharge temperature sensor after the compressor. The fault is detected and isolated before the wash condition of the compressor is triggered, resulting in fault tolerant estimation of compressor health parameters.charge temperature sensor after the compressor. The fault is detected and isolated before the wash condition of the compressor is triggered, resulting in fault tolerant estimation of compressor health parameters.

The supervision of performance in gas turbine applications is crucial in order to achieve: (i) reliable operations, (ii) low heat stress in components, (iii) low fuel consumption, and (iv) efficient overhaul and maintenance. To obtain a good diagnosis of performance it is important to have tests which are based on models with high accuracy. A main contribution is a systematic design procedure to construct a fault detection and isolation (FDI) system for complex nonlinear models. To fulfill the requirement of an automated design procedure, a thermodynamic gas turbine package (GTLib) is developed. Using the GTLib framework, a gas turbine diagnosis model is constructed where component deterioration is introduced. In the design of the test quantities, equations from the developed diagnosis model are carefully selected. These equations are then used to implement a constant gain extended Kalman filter (CGEKF)-based test quantity. The test quantity is used in the FDI-system to supervise the performance and in the controller to estimate the flame temperature. An evaluation is performed using experimental data from a gas turbine site. The case study shows that the designed FDI-system can be used when the decision about a compressor wash is taken. Thus, the proposed model-based design procedure can be considered when an FDI-system of an industrial gas turbine is constructed.

Monitoring of an industrial gas turbine is important since it gives valuable information for the customer about maintenance, performance and process health. The objective of the paper is to develop a monitoring system for an industrial gas turbine application with a model based diagnosis approach. A constant gain extended Kalman observer is developed. The observer compensates for different ambient conditions such as pressure, temperature and relative humidity, due to the amount of water in the atmosphere. The developed observer, extended with seven health parameters, is automatically constructed from the diagnosis model. These health parameters shall capture deviations in some of the gas path performance parameters such as efficiency, mass flow, turbine inlet area and head loss. The constructed observer is evaluated through a simulation study where the ambient conditions are changed. The considered observer capture the change in different ambient conditions nearly perfect. An observer that does not compensate for different ambient conditions gives an error for about 1–2% for the considered health parameters for the given test case. The constructed observer is also evaluated on measurement data from a mechanical drive site. A degradation in efficiency and mass flow for the compressor due to fouling can be seen in the estimations. After the compressor wash is performed, the degradations for the compressor are partially restored by about 2% which can be seen in the considered health parameters.

Monitoring of an industrial gas turbine is important since it gives valuable information for the customer about maintenance, performance and process health. The objective of the paper is to develop a monitoring system for an industrial gas turbine application with a model based diagnosis approach. A constant gain extended Kalman observer is developed. The observer compensates for different ambient conditions such as pressure, temperature and relative humidity, due to the amount of water in the atmosphere. The developed observer, extended with seven health parameters, is automatically constructed from the diagnosis model. These health parameters shall capture deviations in some of the gas path performance parameters such as efficiency, mass flow, turbine inlet area and head loss. The constructed observer is evaluated through a simulation study where the ambient conditions are changed. The considered observer capture the change in different ambient conditions nearly perfect. An observer that does not compensate for different ambient conditions gives an error for about 1-2 % for the considered health parameters for the given test case. The constructed observer is also evaluated on measurement data from a mechanical drive site. A degradation in efficiency and mass flow for the compressor due to fouling can be seen in the estimations. After the compressor wash is performed, the degradations for the compressor are partially restored by about 2 % which can be seen in the considered health parameters.

Today’s need for fuel efficient vehicles, together with increasing engine component complexity, makes optimal control a valuable tool in the process of finding the most fuel efficient control strategies. To efficiently calculate the solution to optimal control problems a gradient based optimization technique is desirable, making continuously differentiable models preferable. Many existing control-oriented Diesel engine models do not fully posses this property, often due to signal saturations or discrete conditions. This paper offers a continuously differentiable, mean value engine model, of a heavy-duty diesel engine equipped with VGT and EGR, suitable for optimal control purposes. The model is developed from an existing, validated, engine model, but adapted to be continuously differentiable and therefore tailored for usage in an optimal control environment. The changes due to the conversion are quantified and presented. Furthermore, it is shown and analyzed how to optimally control the engine in a fuel optimal way under steady-state conditions, and in a time optimal way in a tip-in scenario.

A compressor model is developed. It is capable of representing mass flow and pressure characteristic for three different regions: surge, normal operation as well as for when the compressor acts as a restriction, i.e. having a pressure ratio of less than unity. Different submodels are discussed and methods to parametrize the given model structure are given. Both the parametrization and validation are supported extensively by measured data. Dynamic data sets include measurements from engine and surge test stands. The compressor model is further validated against a database of stationary compressor maps. The proposed model is shown to have good agreement with measured data for all regions, without the need for extensive geometric information or data.

Increasingly stringent emissions legislation combined with consumer performance demand, have created the need for complex automotive engines. The control of these complex system rely heavily on control oriented models. Models capable of describing all operating modes of the systems are beneficial, and the models should be easily parametrized and enable extrapolation. A large database of automotive compressor maps is characterized, and used to develop, validate and automatically parametrize a compressor flow model capable of describing reversed flow, normal operation and choke. Measurement data from both an engine test stand, and a surge test stand, is used to parametrize and validate the surge capability of the model. The model is shown to describe all modes of operation with good performance, and also to be able to extrapolate to small turbo speeds. The extrapolation capability is important, since compressor maps are shown to lack information for low speeds, even though they frequently operate there in an engine installation.

A method for determining turbocharger performance on installations in an engine test bench is developed and investigated. The focus is on the mapping of compressor performance but some attention is also given to the turbine mapping. An analysis of the limits that an engine installation imposes on the reachable points in the compressor map is performed, in particular it shows what corrected flows and pressure ratios can be reached and what these limitations depend on. To be able to span over a larger  region of the corrected flow a throttle before the compressor is suggested and this is also verified in the test bench.

Turbocharger mapping is a time consuming process and there is a need for a systematic process that can be executed automatically. An engine and test cell control structure that can be used to automate and monitor the measurements by controlling the system to the desired operating points is also proposed.

In experiments, used for constructing the compressor speed lines, it is virtually impossible to control the turbocharger to the exact corrected speed that is postulated by the speed line. To overcome this two methods that compensate for the deviation between measured speed and the desired speed are proposed and investigated. Detailed data from a gas stand is used to evaluate the measurements compared to those that are generated in the engine test cell installation. The agreements are generally good but there is more noise in the engine data and there are also some small systematic deviations.

Turbo performance is represented using maps, measured for one set of inlet conditions. Corrections are then applied to scale the performance to other inlet conditions. A turbo compressor for automotive applications experiences large variations in inlet conditions, and the use of two stage charging increases these variations. The variations are the motivation for analyzing the correction quantities and their validity. The corrections reveals a novel surge avoidance strategy, where the result is that a reduction in inlet pressure increases the surge margin for eight maps studied. The method to investigate the applicability of the strategy is general.

An experimental analysis of the applicability of the commonly used correction factors, used when estimating compressor performance for varying inlet conditions, is presented. The experimental campaign uses measurements from an engine test cell and from a gas stand, and shows a small, but clearly measurable trend, with decreasing compressor  pressure ratio for decreasing compressor inlet pressure. A method is  developed, enabling measurements to be analyzed with modified corrections.

An adjusted shaft speed correction quantity is proposed, incorporating also the inlet pressure in the shaft speed correction. The resulting decrease in high altitude engine performance, due to compressor limitations, are quantified and shows a reduction in altitude of 200 – 600 m, for when engine torque has to be reduced to due limited compressor operation.

Increasingly stringent emissions legislation combined with consumer performance demands, have driven the development of downsized engines with complex turbocharger arrangements. To handle the complexity model-based methods have become a standard tool, and these methods need models that are capable of describing all operating modes of the systems. The models should also be easily parametrized and enable extrapolation. Both single and multiple stage turbo systems can operate with a pressure drop over their compressors, both stationary and transient. The focus here is to develop models that can describe centrifugal compressors that operate both in normal region and restriction region from standstill to maximum speed. The modeling results rely on an analysis of 305 automotive compressor maps, whereof five contain measured restriction operation, and two contain measured standstill characteristic. A standstill compressor is shown to choke at a pressure ratio of approximately 0.5, and the corresponding choking corrected mass flow being approximately 50% of the compressor maximum flow capacity. Both choking pressure ratio and flow are then shown to increase with corrected speed, and the choking pressure ratio is shown to occur at pressure ratios larger than unity for higher speeds. Simple empirical models are proposed and shown to be able to describe high flow and pressure ratios down to choking conditions well. A novel compressor flow model is proposed and validated to capture the high flow asymptote well, for speeds from standstill up to maximum.

A compressor model is developed. It is capable of representing mass flow and pressure characteristic for three different regions: surge, normal operation as well as for when the compressor acts as a restriction, i.e. having a pressure ratio of less than unity. Different submodels are discussed and methods to parametrize the given model structure are given. Both the parameterization and validation are supported extensively by measured data. Transient data sets include measurements from engine test stands and a surge test stand. The compressor model is further validated against a data base of stationary compressor maps. The proposed model is shown to have good agreement with measured data for all regions, without the need for extensive geometric information or data.

Surge is a dangerous instability that can occur in compressors. It is avoided using a valve that reduces the compressor pressure. The control of this valve is important for the compressor safety but it also has a direct influence on the acceleration performance. Compressor surge control is investigated by first studying the surge phenomenon in detail. Experimental data from a dynamic compressor flow test bench and surge cycles measured on an engine is used to tune and validate a model capable of describing surge. A concept time to surge is introduced and a sensitivity analysis is performed to isolate the important characteristics that influence surge transients in an engine. It is pointed out that the controller clearly benefits from a feed-forward term due to the small time frames associated with the transition to surge. In the next step this knowledge is used in the design of a novel surge controller. This surge controller is then compared to two other controllers and it is shown that it avoids surge and improves the acceleration performance by delivering both higher engine torque and turbo shaft speed after a gear change.

A mean value engine model of a two-stroke ma-rine diesel engine with EGR that is capable of simulatingduring low load operation is developed. In order to beable to perform low load simulations, a compressor modelcapable of low speed extrapolation is also investigated andparameterized for two different compressors. Moreover, aparameterization procedure to get good parameters for bothstationary and dynamic simulations is described and applied.The model is validated for two engine layouts of the same testengine but with different turbocharger units. The simulationresults show a good agreement with the different measuredsignals, including the oxygen content in the scavengingmanifold.

Downsizing and turbocharging with single or multiple stages has been one of the main solutions to decrease fuel consumption and harmful exhaust emissions, while keeping a sufficient power output. An accurate and reliable control-oriented compressor model can be very helpful during the development phase, as well as for engine calibration, control design, diagnostic purposes or observer design. A complete compressor model consisting of mass flow and efficiency models is developed and motivated. The proposed model is not only able to represent accurately the normal region measured in a compressor map but also it is capable to extrapolate to low compressor speeds. Moreover, the efficiency extrapolation is studied by analyzing the known problem with heat transfer from the hot turbine side, which introduces errors in the measurements done in standard gas stands. Since the parameterization of the model is an important and necessary step in the modeling, a tailored parameterization approach is presented based on Total Least Squares. A standard compressor map is the only data required to parameterize the model. The parameterization is tested with a database of more than 230 compressor maps showing that it can deal well with different compressor sizes and characteristics. Also, general initialization values for the model parameters are provided using the complete database parameterization results. The results show that the model accuracy is good and in general achieves relative errors below one percent. A comparison of the model accuracy for compressor maps with and without heat transfer influence is carried out, showing a similar model accuracy for both cases but better when no heat transfer is present. Furthermore, it is shown that the model is capable to predict the efficiency characteristics at low speed of two compressor maps, measured with near adiabatic conditions.

Large marine two-stroke diesel engines are widely used as propulsion systems for shipping worldwide and are facing stricter NOx emission limits. Exhaust gas recirculation is introduced to these engines to reduce the produced combustion NOx to the allowed levels. Since the current number of engines built with exhaust gas recirculation is low and engine testing is very expensive, a powerful alternative for developing exhaust gas recirculation controllers for such engines is to use control-oriented simulation models. Unfortunately, the same reasons that motivate the use of simulation models also hinder the capacity to obtain sufficient measurement data at different operating points for developing the models. A mean value engine model of a large two-stroke diesel with exhaust gas recirculation that can be simulated faster than real time is presented and validated. An analytic model for the cylinder pressure that captures the effects of changes in the fuel control inputs is also developed and validated with cylinder pressure measurements. A parameterization procedure that deals with the low number of measurement data available is proposed. After the parameterization, the model is shown to capture the stationary operation of the real engine well. The transient prediction capability of the model is also considered satisfactory which is important if the model is to be used for exhaust gas recirculation controller development during transients. Furthermore, the experience gathered while developing the model about essential signals to be measured is summarized, which can be very helpful for future applications of the model. Finally, models for the ship propeller and resistance are also investigated, showing good agreement with the measured ship sailing signals during maneuvers. These models give a complete vessel model and make it possible to simulate various maneuvering scenarios, giving different loading profiles that can be used to investigate the performance of exhaust gas recirculation and other controllers during transients.

A toolbox for parameterizing the ellipse model, that is a control-oriented compressor model, to any given measured compressor map is described in detail in this document. The compressor model has been developed in previous publications and shown to be capable of accurately reproducing the measured data obtained from gas stand measurements, for a wide range of compressors, starting from small automotive applications to large compressors used in marine propulsion. In addition, it has been shown that it is possible to extrapolate both mass flow and efficiency to the unmeasured low speed region of the compressor in a physical way. The parameterization algorithm is based on Total Least Squares (TLS), which is shown here and in previous publications to be a fast and reliable approach to fit the compressor model to the map. The toolbox is implemented in a Matlab Graphical User Interface (GUI) in order to make it easy for the user to parameterize the compressor model. To demonstrate the workflow and ease of use, a complete step-by-step example of how to work with the toolbox is provided. To further facilitate the user in applying the model, the package also provides implementations of the ellipse compressor model both as a Matlab function and as a Simulink block. This way, the user can quickly and reliably use the results of the parameterization process in a desired application, e.g. including the compressor model of a given compressor map in a combustion engine simulation model.

Optimal control of a heavy duty diesel engine with EGR and VGT during transients is investigated. Minimum time and fuel optimal control problems are defined for transients from low to high output torque. A validated diesel engine model is used with minor changes in order to be suitable for the selected solver. The problem is solved for several feasible minimum EGR fractions and smoke-limiter values in order to provide comparisons. The optimization results show that the smoke-limiter has great effect on the transient duration while the required EGR fraction influences the control signals' shape. The fuel optimal control keeps the control actuators more closed than the time optimal, however both time and fuel optimal results become very similar when high EGR fractions and smoke-limiter values are required.

A complete and compact control-oriented compressor model consisting of a mass flow submodel and an efficiency submodel is described. The final application of the model is a complete two-stroke mean value engine model (MVEM) which requires simulating the compressor operating at the low-flow and low-pressure ratio area. The model is based on previous research done for automotive-size compressors, and it is shown to be general enough to adapt well to the characteristics of the marine-size compressors. A physics-based efficiency model allows, together with the mass flow model, extrapolating to low-pressure ratios. The complexity of the model makes its parameterization a difficult task; hence, a method to efficiently estimate the 19 model parameters is proposed. The method computes analytic model gradients and uses them to minimize the orthogonal distances between the modeled speed lines (SpLs) and the measured points. The results of the parameter estimation are tested against nine different standard marine-size maps showing good agreement with the measured data. Furthermore, the results also show the importance of estimating the parameters of the mass flow and efficiency submodels at the same time to obtain an accurate model. The extrapolation capabilities to low-load regions are also tested using low-load measurements from an automotive-size compressor. It is shown that the model follows the measured efficiency trend down to low loads.

A method that finds fuel optimal speed profiles for traveling a predefined distance is presented. The vehicle is modeled using a quasistatic formulation and an optimal control problem is defined. In addition, the solving method is based on a multi-phase optimization algorithm based on dynamic programming. This approach results in lower computational time than solving the problem directly with dynamic programming, it also makes the computational time independent of the travel distance. In addition, the simulation generated data can be used to get the solution to several optimal control problems in parallel that have additional constraints. Further a finite time gear shift model is presented to include the gear selection in the optimization problem. The problem also considers speed losses and fuel consumption during the maneuver. The results presented show the optimal speed and gear profiles to cover a distance, making special emphasis at the acceleration phase, where it is optimal to perform a fast acceleration to engage the highest gear as soon as possible. Finally a proposed application is to use the simulation data to provide eco-driving tips to the driver.

A transmission with both high comfort and high efficiency is the Automated Manual Transmission (AMT). To be able to control and fully utilize this type of transmission it is of great importance to have knowledge about the torque transmissibility curve of the clutch. The transmitted torque in a slipping dry clutch is therefore studied in experiments with a heavy duty truck (HDT). It is shown that the torque characteristic has little or no dependence on slip speed, but that there are two dynamic effects that make the torque vary up to 900 Nm for the same clutch actuator position. Material expansion with temperature can explain both phenomena and a dynamic clutch temperature model that can describe the dynamic torque variations is developed. The dynamic model is validated in experiments, and it is shown to reduce the error in transmitted torque from 7 % to 3 % of the maximum engine torque compared to a static model. Clutch wear is also a dynamic phenomenon that is of interest to track and compensate for, and therefore the model is augmented with an extra state describing wear. An observability analysis is performed showing that the augmented model is fully or partially observable depending on the mode of operation. In particular, by measuring the actuator position the temperature states are observable, both during slipping of the clutch and when it is fully closed. An Extended Kalman Filter (EKF), which observes the temperature states, was developed since it is straight forward to incorporate different modes of operation. The EKF was evaluated on measurement data and the estimated states converged from poor initial values, enabling prediction of the translation of the torque transmissibility curve. The computional complexity of the EKF is low and thus it is suitable for real-time applications. Modeling, parameter estimation, observer design and validation are all carried out using production sensors only and therefore it is straight forward to implement the observer in a production HDT following the presented methodology.

In powertrain analysis, simulation of driveline models are standard tools, where efficient and accurate simulations are important features of the models. One input signal with high impact on the accuracy is the road slope. Here it is found that the amplitude discretization in production road-slope sensors can excite vehicle shuffle dynamics in the model, which is not present in the real vehicle. To overcome this problem road-slope information is analyzed with the aid of both measured and synthetic road profiles, where the latter are generated from regulatory road specifications. The analysis shows that it is possible to separate vehicle shuffle resonances and road-slope information, and designs are proposed for on- and off-line filtering of the road-slope-sensor signal in spatial coordinates. Applying the filter to measured data shows that vehicle shuffle is significantly attenuated, while the shape of the road slope profile is maintained. As a byproduct the use of smoothing the rolling resistance is shown.

A dry clutch model with thermal dynamics is added to a driveline model of a heavy-duty truck equipped with an automated manual transmission. The model captures driveline oscillations and can be used to simulate how different clutch-control strategies affect vehicle performance, drivability and comfort. Parameters are estimated to fit a heavy-duty truck and the complete model is validated with respect to shuffle, speed trajectory, clutch torque and clutch lock-up/break-apart behavior. The model shows good agreement with data. Furthermore the model is used to study the effect of thermal expansion in the clutch on launch control. It is shown that the effect of thermal expansion, even for moderate temperatures, is significant in launch control applications.

The transmitted torque in a slipping dry clutch is studied in experiments with a heavy duty truck. It is shown that the torque characteristic has little or no dependence on slip speed, but that there are two dynamic effects that make the torque vary up to 900 Nm for the same clutch actuator position. Material expansion with temperature can explain both phenomena and a dynamic clutch temperature model with two different time constants is developed. The dynamic model is validated in experiments, with an error of only 3% of the maximum engine torque, and is shown to improve the behavior significantly compared to a static model.

Abstract Development of efficient control algorithms for the control of automatic transmission systems is crucial to maintain passenger comfort and operational life of the transmission components. An optimization framework is developed by state space modeling of a powertrain including a nine speed automatic transmission, diesel engine, torque converter and a model for longitudinal vehicle dynamics considering drive shaft as the only flexibility of the driveline. Emphasis is set on the kinematics of the automatic transmission with the aim of modeling for gearshift optimal control during the inertia phase. Considering the interacting forces between planetary gearsets, clutches and brakes in the transmission, kinematic equations of motion are derived for rotating transmission components enabling to calculate both transmission dynamics and internal forces. The model is then used in optimal control problem formulations for the analysis of optimal control transients in two up-shift cases.

Optimal transients of a hybrid powertrain are calculated with the aim to give a smooth and time efficient acceleration. It is shown that there is a trade-off between time and driveline oscillations where high oscillations can be avoided by slightly longer acceleration time and proper control of the electrical and diesel power sources. During a low oscillation acceleration, there is still the possibility to reduce the amount of total consumed electrical and fuel energy. This is investigated by calculation of optimal controls during acceleration for a fixed time while penalizing the usage of energy in a low oscillation acceleration. The balance between electrical and diesel energy usage during the acceleration is also investigated. The results show that to avoid extreme transients by optimal control, a multidimensional formulation of the objective function including different properties should be considered.

To investigate the optimal controls of a diesel-electric powertrain during a torque controlled gearshift, a powertrain model is developed. A validated diesel-electric model is used as the power source and the transmission dynamics are described by different sets of differential equations during torque phase, synchronization phase and inertia phase of the gearshift. Using the developed model, multi-phase optimal control problems are formulated and solved. The trade-off between gearshift duration and driveline oscillations are calculated and efficient gearshift transients for a diesel-electric and pure diesel powertrain are then compared and analyzed.

Gearshift optimal control of a hybrid powertrain with a lumped/decoupled transmission model and backlash dynamics in the driveline is studied. A model is used for a heavy duty powertrain including a validated mean value diesel engine model with electric generator, transmission dynamics representing the dynamics of the automated manual transmission system and driveshaft flexibilities. Backlash dynamics are also included in the driveline model by introducing a switching function. By applying numerical optimal control methods and dividing the gearshift process into separate phases, optimization problems are solved to investigate the minimum time and low Jerk gearshift transients. The controls are also calculated with fuel penalties added to the minimum Jerk optimization and the transients are analyzed.

The optimal control of wheel loader operation is used in order to investigate the potentials for fuel cost and cycle time minimization during the short loading cycle. The wheel loader is modeled as a nonlinear system with three control inputs and four state variables where a diesel engine generates the power utilized for lifting and traction. The lifting system is modeled considering the limitations in the hydraulics and also the structural constraints. A torque converter is included in the driveline model which introduces nonlinearities into the system and operates in different modes affecting the fuel consumption. The gear shifts during the loading cycle impose a discrete variable into the system and this is taken care of by representing the loading cycle as a multi-phase optimal control problem with constant gearbox gear ratio in each phase. Minimum fuel and minimum time system transients are calculated and analyzed for two alternative cases one where the torque converter is used to stop the vehicle before reaching the reversing point and another where the service brakes are utilized. The optimal control problem is iteratively solved in order to obtain the trade-off between fuel consumption and cycle time for both braking alternatives. It is shown that although the engine operates at lower speeds when the torque converter is used for braking, the fuel consumption increases as higher torques are demanded from the engine during braking. The increase in fuel consumption is higher in faster cycle operations as the vehicle travels at higher speeds and larger torques are required to stop the vehicle. Wheel loader operators tend to use torque converter braking alternative as it is more convenient; however, it accompanies higher fuel consumption which highlights the importance of developing intelligent and easy to use braking systems.

Time and fuel optimal control of an articulated wheel loader is studied during the lift and transport sections of the short loading cycle. A wheel loader model is developed including engine (with turbo dynamics), torque converter, transmission and vehicle kinematics, lifting hydraulics and articulated steering. The modeling is performed with the aim to use the models for formulating and solving optimal control problems. The considered problem is the lift and transport section of the wheel loader that operates in the short loading cycle, with several different load receiver positions, while the considered criteria are minimum time and minimum fuel. The problem is separated into four phases to avoid solving a mixed integer problem imposed by the gearshifting discontinuities. Furthermore, two different load lifting patterns are studied one with the lifting free and one with the lifting performed only in the last 30 % of the transport. The results show that the optimal paths to the load receiver are identical for both minimum time and minimum fuel cycles and do not change when the loading lifting pattern is altered. A power break-down during the wheel loader operation is presented for the selected cycles of normal and delayed lifting where it is shown that the cycle time remains almost unchanged when lifting is delayed while the fuel consumption slightly decreases in minimum time transients.

Abstract A nonlinear wheel loader model with nine states and four control inputs is utilized to study the fuel and time efficient optimal control of wheel loader operation in the short loading cycle. The wheel loader model consists of lifting, steering and powertrain subsystems where the nonlinearity originates from the torque converter in the drivetrain. The short loading cycle, from loading point to a load receiver and back to the loading point, for a fork lifting application is described in terms of boundary conditions of the optimization problem while the operation is divided into several phases with constant gearbox gear ratios in order to avoid discontinuities due to discrete gear ratios. The effect of load receiver standing orientation on the wheel loader trajectory, fuel consumption and cycle time is studied showing that a small deviation from the optimal orientation (≈ 20 [deg]) results in up to 18 % higher fuel consumption in the minimum time cycles. Also, an alternative lifting strategy where for operation safety load is lifted only when wheel loaders moves forward is studied showing that this increases the fuel consumption of a typical 25 [sec] cycle only less than 2 %. The wheel loader path between loading point and load receiver is also calculated by optimization and analyzed for different cases. It is shown that when the load receiver orientation is not optimized and is set manually, the time or fuel optimal paths will differ from the shortest distance path, however when the load receiver orientation is calculated by optimization the fuel, time and shortest distance paths become identical.

Abstract Optimal control of a wheel loader operating in the short loading cycle is studied in order to investigate the potentials for fuel consumption reduction while maintaining acceptable production rates. The wheel loader is modeled as a system with five states and three control inputs including torque converter nonlinearities. The torque converter is modeled with no lockup enabling power transmission in both directions. The geometry of the wheel loader boom and the demanded force in the lift cylinders during lifting are used to ensure that the in-cylinder pressure remains below component’s limits. The lift-transport section of the short loading cycle is divided into four phases due to discontinuities in the gearbox ratios and fuel consumption is calculated in each phase. Time optimal and fuel optimal transients of the system and the power consumption in each and every component is presented showing the dominance of the torque converter losses compared to the other components especially in the time optimal solutions. It is shown that introducing path constraints on the maximum lifting speed of the bucket due to limitations in hydraulic pumping speed moves the diesel engine operation towards higher speeds in order to maintain the lifting speed. Trade-off between fuel optimal and time optimal transients is calculated which is found to be in agreement with the results of experimental studies.