Optimal energy management for long-haul trucks Fuel

Volvo Group Trucks Technology
Gothenburg, Sweden
embotech GmbH
Physikstrasse 3, ETL K10.1
CH-8092 Zurich
http://www.volvogroup.com/
Tel. +41 632 6298
e-mail: [email protected]
Optimal energy management for long-haul trucks
Fuel costs represent a significant proportion of the cost of operation for heavy duty vehicles such as long-haul
trucks. Accordingly, there is strong demand in the marketplace for more fuel efficient trucks and manufacturers
have to develop new fuel saving measures to stay competitive. The consensus is that increasingly holistic approaches to vehicle control have to be taken to realize significant fuel savings. Increasing electrification of longhaul trucks presents a great opportunity for these approaches by combining measures such as regenerative braking, forecasts of the vehicle trajectory, and the larger control influence over separate electrical subsystems. At the
same time, more powerful control hardware enables the use of advanced optimal control methods.
Volvo Group Trucks Technology investigated together with project partners in the CONVENIENT research
project how prediction and integrated control schemes can contribute to more fuel efficient long-haul trucks.
The code generation system FORCES Pro from embotech was used to quickly prototype optimal predictive
controllers that were then implemented on the CONVENIENT prototyping truck.
The Challenge
The engine cooling system is one of the largest parasitic loads in long-haul trucks. The power demand of the
cooling system pumps and fans should be reduced as much as possible by exploiting predictive information
about the vehicle route including location and steepness of future inclines and declines. Additionally, available
energy buffers, e.g. cooling water temperature, vehicle kinetic energy, and battery, should be used to further reduce the overall energy consumption.
At the same time, operating constraints on the engine must not be violated. Examples of such constraints are
the coolant temperature, the battery state-of-charge and the battery voltage which all should be within specified
boundaries.
The Solution
Model predictive control is the natural choice when predictions and constraints have to be incorporated into a
control design. The FORCES Pro code generation system from embotech was used to rapidly prototype an optimal control strategy that minimizes fuel consumption by exploiting knowledge of the vehicle trajectory while
observing all operational constraints.
A linearized, holistic vehicle model was used to enable the optimal controller generated with FORCES Pro to
efficiently exploit the interplay of different vehicle systems such as regenerative braking, battery, and cooling water temperature, for more fuel efficient operation.
The generated solvers were evaluated in simulation on a high-fidelity vehicle model. Since solvers generated with
FORCES Pro are embeddable everywhere, the same control algorithms could immediately be deployed to
dSPACE rapid prototyping hardware without re-implementation for the embedded system.
The Result
The resulting MPC controller was evaluated against
a baseline, state of the art PI-controller and a solution based on dynamic programming that represents
the theoretical maximum in fuel efficiency because it
solves the problem over the whole driving cycle.
The latter solution cannot be implemented in realtime due to its significant computational requirements. The algorithms generated with FORCES Pro,
on the other hand, could be deployed to the
CONVENIENT prototype truck to evaluate the
system in tests.
For two representative drive cycles, the MPC controllers implemented with FORCES Pro achieved a reduction
in the power consumption of the alternator of 16% and 18% over the baseline controller, respectively. The theoretically maximal reduction computed with dynamic programming for the two cycles were 34% and 19%, respectively, indicating that FORCES Pro can realize a major portion of the possible fuel savings. Operational constraints are of course satisfied since they are included in the FORCES Pro design from the beginning.
Despite the more sophisticated control methodology used, development time of MPC controllers with FORCES
Pro is significantly reduced for two main reasons: On the one hand, the systematic, model-based approach to
control design does not necessitate extensive test-drives and cumbersome hand-tuning of control parameters by
experts. On the other hand, the auto code generation capabilities of FORCES Pro let the control engineers focus on the control problem rather than implementation and algorithmic issues, shortening prototyping cycles and
reducing the number of implementation errors that have to be fixed.
Details about the CONVENIENT research project can be found at http://www.convenient-project.eu/ while the
control problem considered here and related work is described in the following publications:
Lindgärde, O., Feng, L., Tenstam, A., and Söderman, M., "Optimal Vehicle Control for Fuel Efficiency," SAE Int. J.
Commer. Veh. 8(2):682-694, 2015, doi:10.4271/2015-01-2875.
Lindgärde, O., Söderman, M., Tenstam, A., and Feng, L., "Optimal Complete Vehicle Control for Fuel Efficiency,"
Transportation Research Procedia 14: 1087-1096, 2016, doi:10.1016/j.trpro.2016.05.179.
Johannesson, L., Murgovski, N., Jonasson, E., Hellgren, J., Egardt, B., "Predictive energy management of hybrid
long-haul trucks", Control Engineering Practice 41: 83-97, August 2015, doi:10.1016/j.conengprac.2015.04.014.