Abstract
Modern thermal management today provides numerous functions to reduce
fuel consumption and to meet the CO2 emission levels of the ACEA of 140 g/km in the year
2008. But increasing the average engine component temperature provides for further
potential.
Currently control elements such as map thermostat valve or electric water pump are used. The
corresponding thermal management software directs the control devices to control the coolant
temperature to a target value by using online data from the engine control unit as input
signals. However, thermal inertia of engine components and fluids as well as transient driving
sometimes prevent from reaching optimum solid temperatures.
The basic idea presented in this paper is to enable the controller to predict operating states of
the engine and vehicle by comparing the anticipated generation of thermal energy with the
existing cooling potential. In order to fulfil this goal the "predictive thermal management
(PTM)" uses expert knowledge of the driver behaviour as well as the road and environment
type. Based on the available information, the controller provides for a cooling power reserve
according to the expected request in steady state mode. In transient mode a prediction of
engine block temperatures is performed by using neuro-physical models in order to
effectively delay a cooling power increase.
In addition to a broad range of performed road and engine tests a thermal simulation model of
the vehicle was developed. It is used to support measurements, to contribute to better
understanding of certain effects and to test strategies of the predictive control.
Keywords - friction loss, thermal management, cooling potential, driver behaviour,
environment type, predictive control