| Abstract (english) | Consumer acceptance of electric vehicles is increasing strongly, with the trend bound to
continue in the future due to beneficial regulations, government incentives, and consumer's
awareness and willingness to shift towards sustainable mobility. Although the innovation in
automotive industry is accelerating and the declared range of current battery electric vehicles
(BEVs) is increasing, their mass market share is still hindered due to long and widely
unavailable charging and end-users’ perception of lacking BEVs range. The already restricted
driving range of BEVs is significantly reduced in extremely hot and cold ambient conditions
due to high energy consumption of the heating, ventilation and air-conditioning (HVAC)
system. To overcome the BEV range reduction in extreme weather conditions, new energyefficient
HVAC systems have been developed recently for improved cabin heating and cooling
efficiency. These are typically vapor-compression cycle-based heat pump systems with
integrated cabin, battery, and powertrain thermal management, and they support operation in
both heating and cooling mode. The advanced BEV HVAC systems are characterized by an
increased number of actuators, which makes the energy management and control system design
more challenging. To minimize the power consumption at a favourable level of thermal
comfort, it is necessary to develop new control systems that can optimally coordinate multiple
and often redundant actuators of the HVAC system, and which utilize optimisation-based
control methods, such as control allocation or model predictive control.
The thesis first presents modelling of an advanced heat pump-based BEV HVAC system and a
cabin thermal dynamics system, which paves the road for model-based optimal control system
design. Next, dynamic programming-based offline control trajectory optimization is carried out
to gain insight into the optimal control actions for various operating conditions and obtain
guidelines for the design of online control systems. Finally, a cascade control strategy based on
the optimal control allocation and a nonlinear model predictive control strategy are designed
for the considered HVAC system. Both control systems are verified in simulation environments,
while the cascade control strategy is also implemented in a B-segment BEV and experimentally
examined in hot and cold weather conditions.
The main aim of the thesis is to design optimal control systems for a passenger cabin heating
and cooling system of an electric vehicle, which coordinate multiple redundant actuators,
accounts for the dynamics and constraints of the overall system and utilizes predictive information such as vehicle's driving cycle and ambient conditions, in order to improve energy
efficiency and maintain high level of thermal comfort in extremely cold and hot weather
conditions.
The thesis is organized in nine chapters, whose content is summarized in what follows.
Chapter 1: Introduction. Outlines the motivation for the presented research and gives a
literature review of the three main topics of the thesis, which are modelling, optimization, and
control of BEV HVAC systems. Finally, it states the main hypothesis and overviews the thesis.
Chapter 2: Functional description of passenger cabin heating and air-conditioning system.
Presents the considered heat pump-based BEV HVAC system. The chapter first describes the
working principle of two main operating modes: heating and cooling. Next, the main feedback
control loops are defined, and the control system design requirements are described, including
the considered thermal comfort index. Finally, two control system concepts, which are designed
in the rest of the thesis, are proposed. The first concept is based on cascade control structure, in
which the superimposed cabin air temperature controller commands the heating/cooling power
to optimal control input allocation algorithm, which transforms the power demand into
references for low-level feedback controllers and auxiliary open-loop control inputs. The
second concept is based on nonlinear model predictive control (NMPC) that regulates the cabin
air temperature and replaces the superimposed cabin air temperature controller and optimal
allocation, while directly setting the references for low-level controllers.
Chapter 3: Modelling of passenger cabin heating and cooling system. Outlines several
simulation models used in the thesis. Detailed physics-based HVAC system model, developed
within a wider project team and implemented in Dymola environment, is used for the purpose
of control system simulation verification, multi-objective optimisation-based control input
allocation design, and low-order models' parametrization. The low-order control-oriented
models are used for the low-level HVAC control system design, control trajectory optimization
and NMPC system design. The low-level HVAC feedback control system design is based on a
linear autoregressive model with exogenous inputs, which describes the cabin inlet air and
superheat temperature transients with respect to compressor speed and electronic expansion
valve control inputs. Next, nonlinear HVAC system models of first and second order are
presented, which describe the low-level controlled cabin inlet air temperature dynamics
including the superheat temperature control loop. Model parameters (time constants and damping ratio) are determined by means of numerical identification procedure, which is based
on detailed physics-based simulation model responses for a large set of operating points. The
obtained model parameter maps are fitted by appropriate analytical functions. Next, nonlinear
regression models of HVAC system power consumption and PMV thermal comfort index are
presented, which are needed for the sake of cost function formulation. Finally, nonlinear singlezone
cabin models of first and second order are presented. The first-order nonlinear cabin model
describes the cabin air temperature transient process, and it is used in control trajectory
optimization, whereas the second-order model additionally describes the cabin body
temperature transient process, and it is used in NMPC system design.
Chapter 4: Control trajectory optimization. Proposes a dynamic programming-based (DP)
method for optimization of HVAC system control trajectories. The HVAC system and cabin
dynamics are represented by the first-order nonlinear models, and the DP algorithm is
implemented in C++ programming language to enhance the computational efficiency. The cost
function reflects the following two conflicting criteria: PMV-based thermal comfort index and
HVAC system energy efficiency. Two approaches of accounting for the energy efficiency are
considered: (i) through maximization of HVAC system coefficient of performance (COP) and
(ii) via minimization of HVAC system electric power consumption. Minimization of the DP
cost function is subject to hard constraints on control variables, as well as constraints that reflect
a limited HVAC operating range. Control trajectory optimization is carried out for winter and
summer ambient conditions, and different cost function setups, thus yielding Pareto optimal
frontiers. The optimization results are analysed with the aim of gaining insights into the optimal
control performance and obtaining guidelines for control system design.
Chapter 5: Optimal control input allocation. Proposes an offline multi-objective genetic
algorithm-based optimization method for generating control input allocation maps. According
to the cascade control concept, the inputs to optimal control allocation are the cooling or heating
power demand, and the cabin air state determined by temperature and relative humidity. The
optimization method relies on detailed physics-based HVAC simulation model, while cabin
model is omitted as cabin air state is reflected by an operating point for which the optimization
is conducted. Firstly, the COP is maximized in both operating modes to obtain optimal control
inputs, which include cabin inlet air temperature reference, blower fan air mass flow, secondary
coolant loop pumps’ speeds and main radiator fan power level. The obtained optimal control
input allocation maps are fitted by proper analytical functions to facilitate implementation and calibration. Additionally, multi-objective optimization is carried out with the aim of
simultaneously minimizing the HVAC power consumption and the thermal comfort index. In
this case, infrared heating panels' (IRP) control inputs are considered, as well. The multiobjective
optimization yields Pareto optimal frontiers, which are analysed with the aim of
gaining insight into potential thermal comfort improvement when utilizing infrared heating
panels and providing guidelines for online thermal comfort control system design.
Chapter 6: Hierarchical control strategy design. The optimal control input allocation maps,
obtained in Chapter 5, are incorporated into a proper cascade control strategy. This chapter first
outlines the design of a superimposed cabin air temperature feedback controller and a PMVbased
feedback controller acting through IRPs. Next, the design of low-level feedback
controllers is presented, including optimization-based design of gain-scheduling maps. The
cascade control system performance is verified through simulations in heat-up and cool-down
scenarios, which start from ambient conditions and last until the thermal comfort is reached.
The impact of various superimposed controller and control allocation setups on energy
consumption and thermal comfort metrics is analysed. Finally, steady-state simulations are
carried out to analyse the extent to which the cabin air temperature reference can be lowered
for reduced power consumption, where the thermal comfort degradation is compensated for by
applying IRPs.
Chapter 7: Nonlinear model predictive control (NMPC). Presents the design of NMPC-based
HVAC system control strategy. First, the optimal control problem is formulated, and it includes
optimization of cabin inlet air temperature and mass flow trajectories on a receding horizon,
which simultaneously minimizes the thermal comfort index and the HVAC system electric
energy consumption. NMPC accounts for the HVAC system and cabin dynamics, a limited
HVAC operating range and predictive information about disturbances, such as vehicle velocity
and ambient air temperature. Next, transformation of the optimal control problem into a
nonlinear program based on the direct multiple shooting method is presented. Finally, the
NMPC system is verified in winter and summer ambient conditions for different cost function
settings, and it is compared with cascade control strategy.
Chapter 8: Experimental verification of cascade control strategy. Presents implementation of
the cascade control strategy, designed in Chapter 6, within an experimental B-segment battery
electric vehicle (BEV). The chapter first describes the experimental vehicle and its control
hardware system, consisting of the main computer, which is used for HVAC system control and human-machine interface communication, and an electronic control unit, which communicates
with actuators and sensors and contains safety features. Then, details of cascade control strategy
implementation within the control hardware are presented, including implementation of
practical modifications, such as safety-related refrigerant pressure controllers and robust HVAC
system start-up procedure. Results of initial commissioning of the control strategy are
presented, which are the basis for additional control strategy calibration. The modified control
strategy is experimentally validated in a climate chamber in hot and cold ambient conditions,
and the obtained performance metrics are analysed.
Chapter 9: Conclusion. Gives the concluding remarks, outlines the possible future work
directions, and states the following main contributions of the doctoral thesis: (i) dynamic
programming-based control trajectory optimization algorithm for a passenger cabin heating and
cooling system of an electric vehicle, which minimizes the electric energy consumption and
provides a high level of passenger thermal comfort; (ii) a cascade control strategy of passenger
cabin heating and cooling system based on a superimposed cabin air temperature controller and
optimal allocation of references for low-level controllers; (iii) an optimal control strategy of
passenger cabin heating and cooling system based on model predictive control, which
coordinates multiple actuators with the aim of increasing vehicle driving range while
maintaining high level of passenger thermal comfort. |
