This article focuses on the use of a free-liquid-piston engine compressor (FPEC) for compact robot power. The FPEC presented in the article combines the engine and the compressor into a single unit. FPEC, a high-power density form of actuation, can help operate human-scale robots. An energy source that provides pneumatic power presents an appealing alternative that alleviates many of the scalability problems of hydraulics while preserving a high actuation power density. The system also presents additional advantages such as power-on-demand with no idle. Taking advantage of the high inertance piston, high-pressure air and high vapor pressure fuel enable the engine to operate in an inject and fire cycle. Dynamically, the FPEC is similar to a bug converter circuit in that the flow is amplified and the high-inheritance piston plays the same energetic role as the inductor. The data suggests that pneumatic systems using the FPEC as a power source would exhibit system energy densities comparable to, if not better than, the best electrochemical systems.
Human-scale robots operating below 1 kW are commonly powered by batteries. Given the low energy density of state-of-the-art rechargeable batteries, operational times of these systems are only on the order of 15 to 30 minutesef0011. The low energy density problem of such systems is often compounded by low power density actuation, resulting in a heavy system that consumes the potential to perform tasks beyond basic mobility. What is needed to break this technological barrier is a system with a high energy density portable power source coupled to a high power density form of actuation.
One such system with an alternative energy source and actuation is seen on Boston Dynamic’s BigDog, which utilizes an internal combustion engine/hydraulic pump and hydraulic actuation system. BigDog’s engine provides about lokW of power for over 2.5 hoursef0022. However, it is difficult to scale this approach down due to the mass of the working fluid as well as components such as return lines, accumulators, control valves, and hosing.
An energy source that provides pneumatic power presents an appealing alternative that alleviates many of the scalability problems of hydraulics while preserving a high actuation power density. Current small-scale engines coupled to compressors present an unreasonably low system efficiency and resulting low energy density. The free-piston engine/compressor (FPEC) presented here combines the engine and the compressor into a single unit. This approach achieves an adequately efficient transduction from chemical potential to pneumatic potential energy by presenting a dynamic load provided by the liquid piston while reducing the overall mass of the conversion system relative to a conventional solution. The system also presents additional advantages such as power-on-demand with no idle. By achieving even a modest efficiency and a reasonable FPEC mass, the high energy density of hydrocarbon fuels (~45 MJ/kg) results in a system energy density many times that of the best batteries (~700 kJ/kg).
Free piston engine compressors operate by separating a combustion chamber from a compressor head with a “free-piston” which moves in response to the pressure forces on either side. Although this compact means of combining the engine and compressor was invented in 1928ef0033, difficulties in control prevented widespread adoption. Controlling the intake and exhaust timings of free-piston engines is challenging because the stroke is not kinematically constrained and the piston position is often difficult to measure. The typical cam-based control seen in non-free-piston engines is therefore not typically available. The FPEC overcomes these challenges by measuring the pressure state in the compressor headef004 ref0054,5.
The liquid piston of the FPEC is constructed by trapping water in a pipe between two elastic membranes (high temperature silicone), thereby combining a mass and spring element in a compact package. This arrangement is shown schematically in Figure 1. The liquid piston’s cross-section is larger at the ends and smaller in the intermediate region, which increases the inertance of the piston while reducing the mass.
Taking advantage of the high inertance piston, high-pressure air and high vapor-pressure fuel enables the engine to operate in an “inject and fire” cycle, which differs significantly from 4-stroke or 2-stroke engines. In this inject-and-fire mode of operation, compressed air and fuel is injected against the wall of the piston. The inertia of the piston presents a high dynamic load such that compression is maintained until injection has finished and the spark can ignite the mixture. The combustion pressure accelerates the water through the pipe, expanding the opposite membrane into the compressor head, forcing air into the high-pressure air reservoir as shown in Fig. 2. The exhaust valve on the combustion head is then opened, the elastic membranes return the piston to its original position, and atmospheric air is breathed into the compressor head. Some of the compressed air is then utilized in the next cycle.
The free-piston engine compressor has progressed through three distinct prototypes as shown in Fig. 3. Each prototype was modeled and then experimentally validatedef0066,-ef01616. The validated model of each version was used to guide and refine the design of the next prototype. The main refinement between prototypes one (Fig. 3a) and two (Fig. 3b) was an increased piston inertance and lower piston mass. The first prototype had a separate combustion chamber that would open to the elastic piston wall in response to combustion pressure. This separated combustion chamber was not necessary in prototype two due to the high dynamic load presented by the inertance to maintain pre-combustion pressure.
The main refinement between the second and third prototype (Fig. 3c) was a figure-8 piston configuration to reduce dynamic reaction forces during operation. Distinct from solid piston engines, the liquid piston enables engine balancing in a single piston engine. Table 1 shows a comparison of computed force, moment, impulse, and angular impulse for each of the three prototype configurations.
Dynamically, the FPEC is similar to a buck converter circuit (shown in Fig. 4) in that the flow is amplified and the high-inertance piston plays the same energetic role as the inductor.
In a buck converter, a high voltage/low current source is converted to a low voltage/high current across a load at a high efficiency using an inductor, diode, and a switch. In the FPEC, more mass leaves the compressor head than enters the combustion head. The admitted air and fuel, when combusted, present a high pressure akin to the source voltage of the buck converter. This source then loads the high inertance piston as is done to the inductor in the buck converter. The kinetic energy stored in the piston is then converted to a higher delivered mass at a lower pressure than the source.
The majority of the experimental results presented below are from the second prototype, which contained more extensive instrumentation in order to validate the dynamic system model and fully energetically characterize the device. Mass and size are best characterized by prototype three. As a self-contained device, prototype three included an on-board microcontroller and all supporting electronics, but contained minimal instrumentation. With regard to prototype two, the measured efficiency (lower heating value of propane fuel to pneumatic potential energy) ranged from 3.45% to 6.63% with an output power range of 9.6 W to 17.9 Wef01616. A virtual-cam control approach applied to this prototype yielded an improved efficiency range of 4.4% to 8.1% with a pressure output range from 380 kPa to 720 kPa (40 to 90 psig) and slightly higher output poweref004 ref0144, 14. Prototype three had a mass of 5.9 kg and was able to deliver compressed air up to 1.2 MPa (157 psig), an improvement over prototype two due largely to model-guided design changes in the compressor head check valve and inertance nozzlesef01717.
Although these efficiencies seem low, it must be remembered that this is the combined efficiency of an internal combustion engine and a compressor, all in a relatively small package. The data suggests that pneumatic systems using the FPEC as a power source would exhibit system energy densities comparable to, if not better than, the best electromechanical systems. The experimentally demonstrated energy density of the FPEC ranges from 2040 kJ/kg to 3750 kJ/kg (resulting stored pneumatic potential energy per kilogram of fuel) compared to about 700 kJ/kg for Li-ion batteries. Accounting for roughly 50% efficiency of the actuation (pneumatic actuators or motors with gear-heads), this translates to a specific work of 1020 kJ/kg to 1875 kJ/kg for the FPEC powered system with pneumatic actuators, compared to about 350 kJ/kg for a Li-ion battery powered system with DC servomotors. Such a threefold advantage, combined with the inherent power-density (and therefore weight) advantages of pneumatic actuators over DC servomotors, devices like the FPEC position pneumatically actuated systems as an attractive option for humanscale, untethered robotic systems. Further design and control optimization of the FPEC is also possible.