Purpose Forklift trucks are generally operated with frequent accelerations and stops, reverse and operations of load handling. This way of operation increases the energy losses and consequently the need for reduction of fuel consumption from forklift customers. This study aims to build a model to replicate the performance of forklifts during real operations and estimate fuel consumption without building a real prototype. Design/methodology/approach AVL Cruise has been used to simulate forklift powertrain and hydraulic circuit. The driving cycles used for this study were in accordance with the standard VDI 2198. Artificial neural networks (ANNs), trained by the results of AVL Cruise simulations, have been used to forecast the fuel consumption for a large set of possible driving cycles. Findings The comparison between simulated and experimental data verified that AVL Cruise model was able to simulate the performance of real forklifts, but the results were only valid for the specified driving cycle. The ANNs, trained by the results of AVL Cruise for a certain number of driving cycles, have been found effective to forecast the fuel consumption of a larger number of driving cycles following the prescriptions of the standard VDI 2198. Originality/value A new method based on ANN, trained by AVL Cruise simulation results, has been introduced to forecast the forklift fuel consumption, reducing the computational time and the cost of experimental tests.

Selective Catalytic Reduction (SCR) technology is currently used to effectively reduce NOx emissions for diesel engines. The present study aims at building a three-dimensional numerical model to evaluate the NOx conversion efficiency and the NH3 slip in an SCR system, reducing the time and resources necessary for design and development process. A numerical model of an SCR system has been built to integrate species transport, heat transfer and flow characteristics along with kinetics of the chemical reactions. In order to systematically simulate situations where the concentration of NH3 at the inlet is not uniform, different NH3 inlet profiles have been built by using a Multivariate Gaussian Distribution and considering the maximum concentration in different locations. The effect of different geometries and NH3 distribution profiles on the NOx conversion efficiency and NH3 slip has been studied. The behavior of the system at different inlet temperatures has been explored and the reaction rates in the monolith have been analyzed. The study has been extended by studying the effect of different NH3/NO ratios to provide a more complete comparison between different designs. The numerical model has been found useful to take into account many aftertreatment system parameters during the design of an SCR system, maximize the NOx conversion efficiency by modifying the NH3/NO ratio while minimizing the NH3 slip, providing a comprehensive tool for the optimization of geometrical characteristics of an SCR system.

Lithium batteries are increasingly used in electric vehicle applications. However, different manufacturing processes and technical constraints lead to battery inconsistency, even for batteries in the same production batch. High-rate discharging negatively affects battery consistency and results in service life reduction. A multi-parameter sorting method at high-rate operation was proposed in this study. The method was applied to sort batteries for cars. The sorted datasets were compared and analyzed by the fuzzy C-mean clustering method, the K-means clustering method, and the simulated annealing genetic algorithm. The comparisons proved that the genetic annealing algorithm was more suitable for battery classification. The clustered batteries were assembled into modules in series and parallel for experimental validation. The test results showed that the battery module cycle life was improved.

A challenging problem in energy storage systems for stand-by application, to obtain the best performance from the lithium cell battery packs/module, is the: • Cells homogeneity, • low and high operative temperature, • cells and module balancing.

The development of automotive exhaust sensors is generally based on a time and resources consuming empirical approach. Building numerical models to study the behavior of the sensors is therefore an important key to reduce the development time and improve sensor quality. Zirconia sensor accuracy and response are dependent from temperature, so temperature is generally controlled in the desired range through a heating element. In the present study an electro-thermo-mechanical model of a heated zirconia oxygen sensor with planar structure has been provided. The numerical model has been used as a tool to study and compare different geometries of heaters. Several heater configurations have been modelled and compared in terms of: temperature rise in different points on the surface of the sensor both on the electrode side and heater side, average and maximum temperature, thermal stress and time for the sensor to be considered functional. The improved heater provided a lower peak temperature, but higher average temperature, more uniform temperature distribution, lower thermal stress and lower time than the base heater to reach the prescribed operational conditions.