Pagina1 STEAM the hydraulic hybrid system for excavators Abstract During the past four years the Institute for Fluid Power Drives and Controls in Aachen has developed a hydraulic hybrid architecture for excavators within the framework of the publically funded STEAM (Steigerung der nergieeffizienz in der Arbeitshydraulik mobiler Arbeitsmaschinen) project. STEAM represents the first holistic approach to improve both the efficiency and performance of hydraulically actuated mobile machinery. Instead of focusing on specific machine subsystems and their losses, the whole machine and the way in which it interacts with its surroundings is analysed in detail. Such an approach leads not only to a decrease in hydraulic losses but also considerably improves the engine s operating efficiency. Motivation For excavators a variety of different hydraulic system architectures are currently available on the market. This is a direct consequence of the wide range of applications such machines is used in. Each task and each operator have their own specific demands regarding precise controllability, system robustness, maintainability, costs and energy efficiency. Depending on the machine class and the daily tasks, one or a number of these properties becomes more important. To meet these requirements OEMs offer several types of Load Sensing or Flow Control systems in single or multi-circuit layouts. Considering the machine as a whole system a general structure can be deduced despite the different hydraulic architectures. Typically an internal combustion engine (ICE) is used as a prime mover, supplying the system with the required mechanical power. One or a number of pumps convert the mechanical into hydraulic power as demanded by the operator. The hydraulic power flow to the individual actuators (rotary and linear drives) is controlled by a series of valves. This system structure leads to a direct coupling between hydraulic power demand and ICE output. The analysis of characteristic duty cycles indicates that the power demand fluctuates considerably, so that the average power demand is significantly lower than the available peak power. Such a system operation implies two causes for a decreased efficiency on the supply side (ICE and pump), which are the transient loading of the diesel engine and furthermore part load operation of these components. Due to the direct coupling between the power demand of the hydraulic actuators and the ICE power, the engine is exposed to transient loading all the time. As a consequence the engine is continuously accelerating or decelerating, which increases fuel consumption and emissions [1]. In typical duty cycles periods of continuous peak power occur rarely. Nevertheless, ICE and pump have to be designed for these operating conditions, if there is no an additional power source. Hence part load conditions are frequent, leading to an inefficient power supply. On the hydraulic side it is a challenging task to distribute the hydraulic power to unequal loaded actuators. In a single circuit layout the pump can only supply one pressure, which has to be the highest load pressure. Thus, the flow to the less loaded actuators must be throttled whereby the system efficiency is reduced even more. These throttling losses can be diminished by utilizing a multi-circuit layout, which increases costs though.
Pagina2 Because of these losses on the supply and the hydraulic side the total system efficiency (useful mechanical work in relation to consumed fuels energy) for an 18t excavator with load sensing hydraulics is consequently in the range of only 8-12 %. A closer look at measurement data (Fig. 1) collected from an 18t excavator performing a 90 truck loading cycle reveals the before mentioned aspects regarding losses. Fig. 1: Distribution of the different system losses The ICE installed in the analysed machine reaches a maximum efficiency of 40 % for steady state operation at an engine speed of approximately 1200 rpm and high torque. This is illustrated in Fig. 1. Even with an ideal supply and hydraulic side, 60 % of the incoming fuel energy will be converted to heat. During the truck loading cycle the motor operates at its rated speed of 1800 rpm to supply rated power. The fluctuating hydraulic power demand as well as the influence of transient loading can be observed. A rapid increase in the hydraulic power demand is followed by considerably higher fuel consumption. Including the losses caused by part load conditions and the power necessary for auxiliary functions (e.g. servo pressure supply, generator, etc.) up to 15 % losses were measured on the supply side. Almost the same amount of losses (13 %) is caused by the valves of the single circuit load sensing system. Another important power flow is the negative part of the actuator power, which represents the power that can be recovered. Energy recovery can be recovered when the boom is lowered or the swing drive is decelerated. The peak negative power (approx. 100 kw) can reach levels similar to the rated engine power. As can be seen in Fig. 1, at the moment
Pagina3 negative power is high it is not possible to reuse it directly to supply other actuators. The only way to make use of this energy is to first store it and distribute it at a later stage. As a result, any new system design must aim to improve engine and pump operation, reduce throttling, and recuperate energy from the actuators. STEAM is based on exactly this design philosophy. STEAM Architecture A simultaneous reduction of both the supply and hydraulic losses can only be achieved with holistic design process. In contrast to other research projects, the first step in designing the STEAM system was to effectively deal with the supply losses. The avoidance of nonstationary, inefficient utilization of ICE and pump is done by decoupling the supply from the actuator power demands. An additional power supply in form of an accumulator is needed. One of the aims of the STEAM project was to use standard off the shelf components. For that reason hydraulic accumulators are used. Compared to electric solutions they possess a higher power density and are easier to maintain. The accumulator assists the ICE and pump when peak power or peak flow is required, thereby the engine has to supply only the average power demand. For this reason the engine speed can be lowered from the standard 1800 to 1200 rpm. In this speed range the motor operates near to its best efficiency point and the speed related losses of the ICE and pump are reduced. Within this accumulator charging circuit, both components undergo only two discrete operating modes. Either the accumulator is being charged using the efficient full load operation of pump and ICE or when the accumulator is fully charged, both components are idling, see Fig. 2. Fig. 2: Accumulator charging circuit [2] Integrating the accumulator into the system also enables automatically energy recovery and storage. As mentioned before this includes both the potential energy while lowering the front attachment and the kinetic energy of the swing drive during braking. With the additional energy source the excavator becomes a hydraulic hybrid. So far the new circuit design improves the supplies efficiency, but such a constant pressure system would lead to immense throttling losses across the valves supplying flow to the unequally loaded actuators. A solution has to be found to match different load pressures to the
Pagina4 accumulator pressure. Before the approach for this dilemma is presented another aim of the STEAM project has to be mentioned. The evaluation of fuel consumption for mobile machinery is a complex task, since many factors can distort the measurement results. To ensure a conclusive system comparison the STEAM system is installed in parallel to the reference system. In this case the benchmark system was a state of the art Linde load sensing system, which is commonly used for an 18t wheeled excavator. As a consequence both systems are operated with the identical engine, single pump supply and actuators, see Fig. 3. Measurements can be conducted with the same machine, the same day and the same surroundings, because a switch between both systems can be done in short time. Hereby the influence of external factors could be minimized. However this approach has some impacts for the STEAM circuit. Fig. 3: Parallel installation of both systems Developing a concept to efficiently distribute power to multiple actuators using a constant pressure system was one the main goals of the research. Supplying the swing drive, a rotary actuator, from a constant pressure system can be realized using a variable displacement motor through the use of secondary control. The motor torque is adjusted in a wide range independently of the supply pressure. This kind of control not only eliminates throttling losses but also allows an efficient recuperation of brake energy. Unfortunately due to the parallel installation of both systems the same valve controlled fixed displacement motor from the existing load sensing system had to be used, see Fig. 4. Depending on the accumulator pressure and the actual load pressure throttling losses across the swing drive are still unavoidable.
Pagina5 Fig. 4: Left: Ideal STEAM circuit for rotary drives, Right: Used layout due to parallel installation with reference system To allow the recovery of the swing brake energy, pressure controlled sequence valves, connected to the accumulator, were installed. Linear actuators are unable to adjust their displacement ratios and cannot adapt their load pressure to the accumulator pressure. This causes considerable throttling. The amount of losses can be reduced by introducing not only one but two discrete pressure levels, High Pressure (HP) and Medium Pressure (MP). In combination with the Tank Pressure level (TP) three different pressure levels are present in the system. With the help of six switching valves each of these pressure levels can be connected to both the piston and rod side end of the cylinders, leading to nine discrete pressure modes or artificial supply pressures. Based on the current actuator load, which is measured with pressure transducers, the machine controller selects the optimal pressure mode. Throttling losses are minimized and flow regeneration or energy recuperation is also enabled. One proportional valve controls the actuator velocity as demanded by the operator. This efficient control layout for linear actuators is illustrated in Fig. 5.
Pagina6 Fig. 5: STEAM circuit for linear actuators [2] The final system layout and the prototype machine (Volvo EW180C) are shown in Fig. 6.
Pagina7 Fig. 6: Final system layout & prototype machine Test Results The first test cycle was chosen to investigate the influence of downspeed engine operation on fuel consumption. The swing drive is ideally suited to this task as both STEAM and the load sensing system use the same swing valve and hydraulic motor to rotate the superstructure. In this case, the only difference between both systems is the supply. The reference system uses the standard pressure regulated load sensing pump at an engine speed of 1800 rpm. STEAM supplies the swing with oil from the high pressure accumulator, which is charged using the pump with the engine running at 1200 rpm, see Fig. 4 [3]. As shown in Fig. 7, the machine s superstructure was rotated 180 back and forth a total of forty times. Fig. 7: Swing Test
Pagina8 The resulting engine operating points for both systems are plotted in Fig. 8. The heights of the individual columns indicate how frequently the engine operates at a certain speed and torque combination. The engine s specific fuel consumption is also plotted in the figure, the lowest value of 205 g/kwh is located around 1000 rpm and at high torque just under the full load line. The load sensing system operates far from this point. STEAM, on the other hand, operates in the region of high efficiency when charging the accumulator. When the accumulator is fully charged the pump swivels back and the engine operates at idle. Fig. 8: Swing test - ICE operating points Pump operation is shown in Fig. 9. The load sensing system shows scattered operating points, while for STEAM the pump operates at one distinct pressure when charging the accumulator. Measurements show that the improved supply architecture results in a 25% reduction in fuel consumption. Enabling swing brake energy recuperation delivers a further 5% reduction. In total, a 30% reduction in fuel consumption was achieved.
Pagina9 Fig. 9: Swing test - Pump operating points Dig and Dump The most important test conducted was a 90 truck loading cycle. This aggressive cycle with full speed actuator motion was chosen to test the limits of the hybrid. In doing so peak power demands occur more often and the capability of the STEAM system at the reduced engine speed could be proven. Fig. 10 shows the test environment. During this loading cycle all the actuators (linear and rotary) are used by the operator simultaneously. The loading movement was repeated 15 times, so that at the end of the cycle approximately 14 t of gravel were dumped into the back of the truck.
Pagina10 Fig. 10: 90 truck loading cycle Peak power is required during boom up operation and the simultaneous accelerating of the swing drive towards the truck. The stored energy in the accumulator assists ICE and pump in such a way that the cycle time remains the same or can even be reduced. Although average power demand is higher than the maximal available power at the reduced engine speed a continuous operation can be achieved. This is possible due to energy recuperation and flow regeneration modes. Pressure modes are also switched during actuator movement. In this case both the arm and bucket actuator begin to extend with a regeneration mode. During the digging phase the external force and consequently the cylinders load pressure increases. A mode switch has to be carried through to prevent the cylinder from losing speed. The ICE operating points for the 90 truck loading cycle are similar to those of the Swing test, see Fig. 11. Using STEAM the engine experiences the desired low speed high torque operation. In contrast the load sensing system operates at the higher engine speed with a more equal distribution along different torque values. In comparison to the swing test idling phases are very short for STEAM due to the high power demand of this cycle. Fig. 11: 90 truck loading cycle - ICE operating points [2] Pump operating points indicate the same differences between both systems. Due to the direct coupling between the supply side and the hydraulic power demand the points for load sensing are scattered across the whole engine s torque range. The STEAM system shows two discrete peaks representing the charging of both the medium and the high pressure accumulator.
Pagina11 Fig. 11: 90 truck loading cycle - Pump operating points [2] Despite the lower engine speed, STEAM achieved the same cycle time and consumed 27 % less fuel. Measurements show that the improved engine operation is responsible for approximately half the improvement. The rest is a direct consequence of the reduced throttling and energy recovery in the hydraulics. These results underline the importance of the holistic approach
Pagina12 However, a closer look at the measurement data reveals losses within the STEAM system, which have prevented even more fuel savings. These losses are a direct consequence of the parallel installation of both systems, which prohibited the installation of a variable displacement motor in secondary control for the swing drive. Therefore the swing motor valve (Fig. 4) is used to adjust the accumulator pressure to the current load pressure leading to significant throttling losses, see Fig. 12. Fig. 12: Comparison of throttling losses at the swing drive Due to space restrictions, compromises had to be made regarding component and piping sizes. This became evident especially during medium pressure charging. The pressure drop between pump and accumulator amounts to 35 bar. Related to the accumulator pressure of approximately 150 bar over 20 % of the charging power is dissipated as heat. Fig. 10: Throttling losses during medium pressure charging
Pagina13 Conclusion The STEAM system is based on a detailed analysis of losses occurring in today s machines. Power losses can be divided into supply losses and hydraulic losses. Supply losses are caused on the one hand by transient loading of the motor and on the other hand inefficient part load operation of the ICE and pump. Hydraulic losses are generated by throttling across the power distributing control valves. Both loss mechanisms are similar in magnitude and cannot be neglected. Thus the holistic approach presented in this abstract may be a possible way to design more efficient hydraulic systems for excavators. First test results are promising. Although engine speed is decreased from 1800 to 1200 rpm the STEAM system achieves the same productivity. Initial field tests show fuel savings of up to 30 % keeping in mind that there is still room for improvement. Acknowledgements Many thanks go to all the individuals who participated in the STEAM project. First and foremost, the German Federal Ministry of Education and Research (BMBF) without whose funding, within the framework of the VIP-Program, the team of researchers at IFAS would not have been able to conduct their work. Special thanks also goes to the VDMA and all the companies that were members of the STEAM advisory board. Finally, thanks goes to Volvo Construction Equipment, especially to the colleagues Ulrich Faß and Heiko Steinbach, for not only providing the test excavator but also for their valuable input and support throughout the project. Literature [1] Lindgren, M.; Hansson, P.-A.: Effects of Transient Conditions on Exhaust Emissions from two Non-road Diesel Engines. In: Biosystems Engineering 87 (2004), Nr. 1, S. 57-66 [2] Leifeld, R.; Vukovic, M.; Murrenhoff, H.: The Road to Efficiency. In ATZ offhighway 3-2016, S. 46-51 [3] Vukovic, M.; Leifeld, R.; Murrenhoff, H.: STEAM - ein hydraulisches Hybridsystem für Bagger. 5th Baumaschinenfachtagung, Dresden, 2015