Optimized Energy Management for 48V Mild Hybrid Drives

48 Volt

6. August 2019 | Engineering Service

When looking at the current drive developments and market forecasts, 48V technology is gaining considerable significance in the automotive industry. This technology is an important part of many automotive manufacturer’s electrification strategies. With moderate technical effort, vehicle fleet CO2 reductions can be achieved in the short term. At the same time, 48V electrification offers significant potential for the reduction of emissions in real operation (real driving emissions – RDE). Given the many functions, such as brake energy regeneration, load point optimization, engine stop sailing, as well as electrification options for charging, driving dynamics, air conditioning, and exhaust systems, it is already foreseeable that the performance and energy reserves of competitive 48V systems will be limited.

The comparison with high-voltage hybrid systems in Figure 1 demonstrates that the operating range of 48V mild hybrid ­systems is clearly moving toward the system limits. The growing number of 48V components additionally increases the dynamics of torque requirements and the variances in terms of operating strategy. This comes with interactions, dynamic framework conditions, and a high system complexity that stretch rule-based operating strategies to their limits. The use of predictive energy management is very promising, since the available electrical energy and power is ideally distributed within the 48V on-board circuit, allowing for ideal operation of 48V systems designed to save costs and resources.

Fig. 1: Comparison of the operating ranges of a high-voltage plug-in hybrid electric vehicle (PHEV) and a 48V mild hybrid electric vehicle (MHEV) in WLTC

Concept vehicle

In cooperation with RWTH Aachen University, FEV has developed a 48V mild hybrid concept vehicle. The vehicle is based on a Mercedes-Benz AMG A45 equipped with all-wheel drive and a seven-speed dual clutch transmission. The series vehicle is equipped with a turbocharged 2.0 l gasoline engine that has a specific output of 133 kW/l. This impressive output is achieved through the use of a large exhaust turbocharger (ETC) that, despite twin-scroll technology, significantly limits the maximum torque in the lower engine speed range and results in a noticeably delayed response. In this context, electrified charging and/or electric torque support can significantly improve elasticity, especially in the economical, lower speed range. The 48V mild hybrid powertrain is schematically represented in Figure 2. The central element is the belt starter generator (BSG) in the belt drive of the combustion engine (CE). The P0 topology enables a variety of hybrid functions such as regeneration, load point shifting, and electric torque support. Since the maximum power that can be transmitted with the belt is limited and there is a permanent connection to the combustion engine, the system is not intended solely for electric driving.

Fig. 2: 48V mild hybrid powertrain of the concept vehicle

There is also an electric compressor (EC) positioned in the charge air path, upstream of the intercooler. The EC reaches a maximum pressure ratio of 1.45 and can significantly increase the charge pressure, and thus the response behavior, in operating ranges with low exhaust enthalpy, regardless of the operating condition of the BSG. The concept vehicle is operated using a rapid control prototyping (RCP) development control device.

Rule-based operating strategy

A driving performance-oriented, rule-based operating strategy with priority-based power distribution controls the electric charging, as well as the electric torque support of the BSG (Figure 3). The operating strategy is made up of the torque-supporting functions in drive management and the overarching power distribution in electric power management. The electric charging is controlled through the pressure ratio between the desired and the current charge pressure in the intake manifold. As long as the waste gate (WG)-regulated ETC does not provide the desired charge pressure, the pressure is additionally increased in the air path through the EC. The required rotational speed is calculated using the compressor diagram of the EC and then limited in accordance with the available electric power.

Fig. 3: Driving performance-oriented rule-based operating strategy with priority-based power distribution

In contrast to electric charging, during which the drive power results from the additional air and fuel mass, the BSG directly converts electric energy into mechanical drive power that supports the combustion engine (Figure 2). The torque required by the BSG results from the difference between the current torque of the combustion engine and the driver’s needs. When the accelerator pedal is pushed, this difference is positive, so that the BSG temporarily replenishes the torque deficit. The BSG torque is then limited in accordance with the available electric power.

The electric power limits of the various individual 48V components are prescribed by the electric energy management. During an acceleration, the 48V battery must also power the cooling agent pump and the 12V system via the DC/DC converter, in addition to the EC and the BSG. It is therefore necessary to carry out a situation-based prioritization of the 48V components. The available battery discharge capacity is, in this context, prescribed by the battery management system (BMS). The available electric discharge capacity for the respective 48V components is then calculated depending on their priority and the actual power consumption of elements with a higher priority. In order to ensure reliable driving operation, the engine cooling and the 12V system have a high priority in this context. The remaining power is made available for the EC and the BSG in consideration of a calibratable power ratio.

Even though such rule-based approaches can be improved through further dependencies, there are principle-based disadvantages. For instance, the operating strategy merely reacts to the current system status and adjusts the parameters regardless of the expected load status. Since, however, the temporal behavior of torque build-up and the efficiency heavily depend on the load status, the selected operating strategy of the electrified drive (CE with ETC, EC, and BSG), and the electric system limits, this control is usually suboptimal.

Optimized Energy Management

Predictive optimization-based energy management strategies use dynamic route information from the electronic horizon for the long-term optimization of route guidance and the speed trajectory. Based on this information and adequate vehicle sensor systems for surroundings detection, hybrid management considers the electric power limits and load prediction to determine ideal trajectories for gear selection, drive torque, and charge strategies for a medium-term horizon. The predicted system values also enable the derivation of an expectable charge condition evolution of the electric energy accumulator, which adapts an energy weighting factor. This factor represents the importance of electric energy in the energy balance sheet and directly influences energy optimization in drive management (Equation 1).

Equation 1

ETot = ∑N k=0E Chem(kT) + ξE El(kT)

At the same time, the response behavior through the regulation of the drive torque, which is made up from the combustion engine torque and the electric torque (Equation 2), is optimized while complying
with the dynamic system limits of the ­ 48V system.

Equation 2

ΔMAntrieb = ∑N k=0M Antrieb, Soll (kT) − MVM(kT) − iRiemenMRSG(kT)

Nonlinear model predictive control (NMPC) relies on a real time-capable, simplified process model of the 48V mild hybrid powertrain. It works with a time horizon of a few seconds and includes time increments of hundredths or tenths of a second for the representation of the nonlinear system dynamics.

The NMPC will calculate the ideal parameter evolution for the WG and the EC, which influence the combustion engine torque through the air path, as well as the torque of the BSG, which can obtained through the addition of the belt drive. This way, both the differences in the temporal behavior of the charge air path and of the BSG torque and their impact on the overall efficiency of the electrified powertrain are taken into account in the optimization.


The NMPC was more closely examined during a validated co-simulation of a B-segment 48V mild hybrid with turbocharged gasoline, electric compression, and P0 BSG. Figure 4 shows a comparison of the NMPC and the rule-based approach for a full-load run-up for various energy weighting factors ξ. An energy weighting factor of four is equivalent to an overall charging efficiency factor of 25 percent, while the electric energy in the limit case of zero, e.g. due to a high battery state of charge and an upcoming downhill drive, is free of charge. Due to the lack of forecasting, the rule-based operating strategy reacts identically in both cases, while the NMPC adjusts the parameters for the WG, the EC, and the BSG based on the situation in order to achieve a desired drive torque. Beyond that, the variation of the optimization parameters shows that the NMPC reduces the drive torque with increasing weighting of energy (h˜NMPR ↑), in order to reduce energy consumption. If the electric energy is free of charge (ξ=0), the drive torque is shifted to the BSG, while the EC builds up charge pressure with the WG open, in order to reduce charge change losses. In contrast to this, at ξ = 4, the NMPC only briefly provides support through the BSG in order to utilize the rapid dynamics of the electric machine and subsequently save electric energy.

Fig. 4: NMPC optimization for various energy weighting factors for a full-load run-up in 5th gear compared to a rule-based approach (weighting ratio energy/response behavior)

The operating strategy, in such an acceleration scenario, is always a compromise between response behavior and energy efficiency. The response behavior is described through the acceleration time and the energy savings through the inverse of the effective drive efficiency. With a variation of the electric power limitation, the framework conditions are changed.

Additionally, for each of these power trajectories, the prioritization of the rule-based strategy and the weighting ratio of the NMPC optimization were varied. It becomes clear that increasing energy savings are at the expense of the response behavior. However, the NMPC resolves the conflict of objectives significantly better and can describe both the energy consumption and the energy savings through the inverse of the effective drive efficiency. With a variation of the electric power limitation, the framework conditions are changed. The stronger the electric power limitation and the smaller the focus on the response behavior, the more the potential of the NMPC develops.

For more information about 48V mild hybrid drives visit 48v.fev.com

Philip Griefnow, RWTH Aachen University
Prof. Jakob Andert, RWTH Aachen University
Dr. Georg Birmes, FEV Europe GmbH