Demonstration of 10-Wp micro organic Rankine cycle generator for low-grade heat recovery

Noboru Yamadaa∗, Yoshihito Tominagab, Takanori Yoshidab

a

Graduate School of Energy and Environment Science, Nagaoka University of Technology, 1603-1

Kamitomioka, Nagaoka, Niigata 940-2188, Japan b

Department of Mechanical Engineering, Nagaoka University of Technology, 1603-1 Kamitomioka,

Nagaoka, Niigata 940-2188, Japan

This document is the Post-print version of a Published Work that appeared in final form in Energy. It should be cited as “Noboru Yamada, Yoshihito Tominaga, Takanori Yoshida, Demonstration of 10-Wp micro organic Rankine cycle generator for low-grade heat recovery, Energy, Vol.78 (15, November 2014), pp.806-813”. To access the final edited and published work see: http://dx.doi.org/10.1016/j.energy.2014.10.075

∗

Corresponding Author. E-mail: [email protected], Phone & Fax: +81-258-47-9762.

1

Abstract In this study, a micro organic Rankine cycle (μ-ORC) generator that produces net positive power from low-grade heat (i.e., temperature below 100°C) was experimentally demonstrated. First, the basic performance of a 10-Wp μ-ORC prototype with a small originally developed displacement-type expander and state-of-the-art commercial direct current (DC) micro-pump was tested. The prototype μ-ORC was coupled with a DC generator to produce a net electrical power of 5 W without an external electricity supply. Finally, the components of the μ-ORC generator were packaged into a compact box to achieve a power density of 214 mW/L (volume), 293 mW/kg (weight), and 3.33 mW/cm2 (footprint), which are the best results achieved in the present study. The maximum achieved isentropic efficiencies of the expander and pump were 4.55% and 30.0%, respectively. There is much scope for further improvement of the power density. The present results suggest that a downscaled ORC generator could be used to recover the energy of widely distributed low-grade heat sources.

Keywords: Rankine cycle, organic Rankine cycle, waste heat recovery, expander, pump, power generation, energy conversion

2

1. Introduction An organic Rankine cycle (ORC) is considered a promising technique for converting hitherto ineffectively utilized low-grade heat into useful mechanical or electrical power. Recently, studies have intensively focused on the development and installation of ORC as a means to mitigate global warming and the energy crisis [1–4]. For example, after the Fukushima Daiichi nuclear disaster in 2011, shifting from large-scale centralized power plants to safer and smaller-scale decentralized power generation has become an emerging issue in Japan. Several ORC generators with power outputs of 1–100 kW have been commercialized over the last few years. The total amount of utilizable low-grade heat is very large; however, the heat sources are widely distributed, that is, each heat source emits a small amount of heat. Thus, the power output range and system size of the ORC generator should be small to accommodate various distributed heat sources, including not only natural energy sources such as geothermal, solar, and biomass but also low-grade heat from factories, combustion engines, electronic devices, environments, and biological bodies. Therefore, reducing the size of the ORC will help expand its applicable range. Many studies have focused on small-scale ORC systems (i.e., power output of 1–10 kW). Volumetric (i.e., displacement-type) expanders such as the scroll [5–17], screw [18,19], gear [20], vane [21], and piston expanders [22] have been used for 1-kW–class systems. Velocity-type expanders such as turbines [23–27] have been used for 10-kW–class and larger systems. Previously used expanders have already been reported [3,28]. However, few reports have presented experimental results for very small ORC generators with potential power outputs of ~100 W or less and the power density for the packaged modules of such generators. In this study, we experimentally demonstrated a packaged 10-W–class ORC generator to investigate the potential of ORC for very small power ranges. A micro ORC (µ-ORC) prototype generator with an originally developed expander and state-of-the-art commercial direct current (DC) micropump was tested at heat source temperatures below 100°C to verify whether it can

3

produce net positive electrical power and operate a working fluid pump without an external electrical power supply. The power density and losses as well as technical issues of the µ-ORC generator for further improvement were considered in detail.

2. Experimental setup and methodology 2.1 µ-ORC without generator The µ-ORC experimental apparatus was developed to be as small as possible. Figs. 1 and 2 respectively show its schematic representation and photograph. Fig. 1(a) shows the piping diagram and measurement points for temperature T and pressure p of the working fluid and Fig. 1(b), the corresponding theoretical pressure–enthalpy (p–h) diagram. Table 1 lists the specifications of the experimental apparatus. HFC245fa was chosen as the working fluid because it is easy to safely handle in the laboratory and has theoretically superior cycle efficiency under the present experimental conditions. Water with an evaporator inlet temperature Thsi = 90°C was used as the heat source; it was heated using an electric heater and circulated throughout the hot-side heat exchanger (i.e., evaporator). Similarly, water with a condenser inlet temperature Tcsi = 15°C was used as the cold source; it was cooled using a chiller and circulated throughout the cold-side heat exchanger (i.e., condenser). Therefore, the temperature difference between the heat and the cold sources (∆T = Thsi − Tcsi) was ∆T = 75°C. The actual average temperatures of both sources fluctuated within ±1°C. The flow rates of the heat and cold source fluids were kept constant during all experiments; they are summarized in Table 2. Brazed-plate heat exchangers were used for both the evaporator and the condenser because of their compactness and high heat transfer performance. The filling mass of the working fluid was ~0.1 kg including the lubrication oil (refrigerant oil). The lubrication oil was added into the closed piping circuit such that the ratio of the lubrication oil to the R245fa purity was 3 wt%. The

4

expander was connected with a torque/speed meter and electric brake to measure its mechanical power. Figs. 3 and 4 show the two displacement-type expanders (i.e., scroll and trochoidal) that were originally developed and integrated into the system; these were selected because displacement-type expanders are highly efficient for small power outputs with minimal components. We applied new expander designs to lower the mechanical loss. The developed scroll expander comprises a non-end plate follower scroll sandwiched between a drive scroll and drive scroll receptor. This simple design eliminates the axial load caused by inner pressure because the drive scroll and drive scroll receptor are joined to each other to contain the inner pressure. The developed trochoidal expander comprises an inner rotor and outer rotor with a needle roller bearing to reduce friction loss between the outer rotor and the casing. As described in Table 1, the designed expansion ratio (built-in expansion ratio) of both expanders was ~2. The designed suction volume per rotation had a similar value of 1.8–1.9 cc/rev for both expanders. To prevent the leakage of working fluid along the output shaft, the same oil seal was used for both expanders. For the working fluid pump, a commercially available, state-of-the-art DC microdiaphragm pump was used. It is important to select an appropriate working fluid pump when constructing the ORC to enable successful operation at low power outputs and temperatures. Thus far, it has been difficult to find commercially available pumps that produce a high discharge pressure of ~1 MPa at the required flow rate with low electricity consumption (i.e., better pump efficiency). The abovementioned DC microdiaphragm pump was expected to satisfy these conditions. It is noted that the pressure fluctuation caused by the DC diaphragm pump did not affect the expander inlet pressure because its amplitude was small and it became negligible in the evaporator.

5

Before the measurements were performed, the ORC process was repeated several times to attain a steady-state condition. The temperature was measured using type-T thermocouples with an uncertainty of ±1.0°C and response speed of ~0.36 s for a range of 20–100°C. The pressure was measured using pressure transducers with an uncertainty of 0.5% for a full-scale load and a response speed of ~5 ms. All measured data were recorded and monitored using a data-acquisition PC with a speed of one datum per second. The estimated measurement uncertainty of the expander mechanical power was within ±5%. The enthalpies and other thermodynamic properties were calculated from the measured pressure and temperature data by using REFPROP ver. 9 developed by NIST [29]. To confirm the repeatability of the experiment, each experiment was repeated at least three times, and a few deviations less than the measurement uncertainty were observed.

2.2 µ-ORC with DC generator An additional experiment was performed in which the µ-ORC was combined with a generator to demonstrate electricity generation. The electric brake was replaced with a DC motor to act as the DC generator. A commercial DC motor with a nominal electrical power of 15 W and nominal no-load rotation speed of 2750 min−1 was chosen and installed as shown in Fig. 5. The generated electrical power was measured using a voltmeter for varied electrical loads. The estimated measurement uncertainty of the generated electrical power was within ±5%. The generator efficiency was estimated as the ratio of the generated electrical power to the mechanical power, which was measured using a torque/rotation speed meter. In this experiment, the trochoidal expander was used because it showed better performance than the scroll expander in the previous experiment.

6

2.3 Packaged µ-ORC generator After measuring the generated electricity, generator efficiency, and net electrical power for the abovementioned unpackaged µ-ORC generator, we attempted to pack all of the µ-ORC components as compactly as possible, as shown in Fig. 6(a), to evaluate the power density. This experiment was aimed at evaluating the power density of the present µ-ORC generator package without an external power supply for comparison with other power generation devices and systems such as thermoelectric generators that basically do not require a permanent electrical power supply. To minimize the system size, the torque/rotation speed meter was uninstalled and thermocouples and related piping joints were removed (pressure sensors were retained). A polystyrene foam box (385 mm × 320 mm × 275 mm) was used as a packing box. In addition, we eliminated the external electrical power supply for the working fluid pump because the µ-ORC generator will be used to generate electricity when there are no permanent electrical power sources around the user’s location. Fig. 6(b) shows the electrical circuit used to drive the working fluid pump with the electricity generated by the DC generator; in other words, the pump is driven by some of the electricity generated by the µ-ORC generator itself. The DC micropump was connected with the DC generator and electrical load in parallel. A capacitor was also installed in parallel to reduce the fluctuation of the generated electricity. The electrical current flow through the circuit I was measured by the voltage V1 at shunt resistance RS1. The electrical current flow through the pump Ip was measured by the voltage V2 at shunt resistance RS2. The current flow through the electrical load R was approximated as I − Ip under the assumption that the current flow through the capacitor was small and negligible.

7

3. Results and discussions 3.1 µ-ORC without generator Figs. 7 and 8 show the measured torque and mechanical power of the scroll and trochoidal expanders in the unpackaged µ-ORC without a generator, respectively. Both expanders showed linear trends for the torque and a peaked curve for mechanical power with respect to the rotation speed. The peak mechanical power of both expanders was similar: 7.2 W at 1500 min-1 (rpm) and 8.4 W at 2000 rpm for the scroll and trochoidal expanders, respectively. During the experiment, the DC micropump had a constant electrical consumption of ~5 W. Consequently, a net positive power (=expander mechanical power - pump electricity consumption) was expected for both expanders. Table 3 summarizes the experimental results for the unpackaged µ-ORC without a generator. The heat exchanger efficiency was high, being at least 90%, indicating that only ~10% of the heat released by the heat source fluid was lost from the heat exchanger to the ambient. The pump isentropic efficiency, defined as the ratio of theoretical power based on an adiabatic pumping process to the measured pump’s electricity consumption, was 30%–40%. These values are slightly high relative to larger scale (kW-class) ORCs, which have been reported to have efficiencies below 25% [3]. In contrast, the expander isentropic efficiencies of the scroll and trochoidal expanders were rather low at 3.57% and 4.55%, respectively. Here, the expander isentropic efficiency is defined as the ratio of the measured mechanical power to the theoretical mechanical power, that is, the power when the working fluid expands adiabatically from the measured expander inlet condition to the outlet condition determined by the working fluid’s condensation pressure at Tcsi = 15°C. The back work ratio (BWR) [3], defined as the ratio of the pump electrical consumption to the expander mechanical power, was estimated to be ~0.7. Clearly, the expander performance was the limiting factor for the present µ-ORC.

8

To clarify the cause of low expander efficiency, two evaluation factors were introduced. The invalid expansion factor was defined as the ratio of the measured expansion ratio (pH/pL) to the expansion ratio for the adiabatic case in which the expander produces the theoretical mechanical power. This factor represents the degree of mismatch between the actual and the designed expansions. A larger invalid expansion factor means that a greater amount of working fluid expanded outside (or inside) the expander without being converted to mechanical power. The filling factor [3,14] was defined as the ratio of the actual mass flow rate to the designed one. Here, the designed mass flow was calculated from the product of the designed swept volume per rotation and the actual rotation speed. The actual mass flow rate was estimated from the pump’s performance curve provided by the manufacturer based on the measured pressure difference between the pump inlet and the outlet. The filling factor represents the degree of leakage loss. A larger leakage loss means that a greater amount of working fluid leaked through the gaps inside the expander without being converted to mechanical power. As summarized in Table 3, the invalid expansion factors of the scroll and trochoidal expanders were 0.19 and 0.22, respectively. Obviously, the built-in expansion ratio of the present expander was too low for the present experimental condition, for which maximizing the built-in expansion ratio or using a piston-type expander that can achieve high expansion ratio can be solutions. The filling factors of the scroll and trochoidal expanders were 2.87 and 2.50, respectively, for Thsi = 90°C. For both expanders, leakage of the working fluid was the dominant cause of low expander efficiency; the leakage can be reduced by increasing the rotation speed of the expander, that is, by modifying the suction volume per revolution (cc/rev) of the expander to be smaller than in the present design. Precise processing of the contacting surfaces and gaps inside the expander is required for increasing the expander rotation speed. In addition, non-negligible mechanical loss may also be caused by friction at the contacting surfaces inside the expander and oil seal on the output shaft, although this was not experimentally measured.

9

3.2 µ-ORC with DC generator Fig. 9 shows the measured expander mechanical power, generated electrical power, generator efficiency, and net electrical power of the µ-ORC generator (unpackaged, with external electrical power supply for pump). The trend of the performance curve was similar to that of the previous experiment. A peak electrical power of 6.2 W, generator efficiency of 83%, and net electrical power of ~1 W were obtained after around 2200 min−1.

3.3 Packaged µ-ORC generator Fig. 10(a) shows the net electrical power of the packaged µ-ORC generator with a self-driven pump. The peak net electrical power of 4.1 W was obtained at an electrical load of 50 Ω. The inset of the figure shows the lighting of a mini-filament lamp installed as the electrical load. The present µ-ORC generator package successfully demonstrated net positive electrical power generation from low-grade heat even though the net power remained low. In this experiment, before ORC operation started, a certain amount of working fluid was moved to the evaporator side in advance and heated when the circulation of the heat source fluid was started to drive the expander. In practice, this condition can be met by manual pumping or by temporarily driving the DC pump with a supplemental power supply such as a dry-cell buttery within the µ-ORC generator package, similar to automotive engines that use a starter motor with a built-in battery. Fig. 10(b) shows the time variation in the expander rotation speed for 10 and 50 Ω. At the start of operation, when there was no electrical load, the expander rotated at ~2800 min−1. When the electrical load was 50 Ω and switch A of the circuit (Fig. 8(b)) was turned ON, the rotation speed decreased to 2200 min−1, and the peak electrical power was generated. When switch A was turned OFF, the rotation speed again became 2800 min−1. Thus, the self-driven µ-ORC generator worked successfully. Conversely, with an electrical load of 10 Ω, the rotation speed 10

decreased steeply because of the high mechanical load, which simultaneously caused a lack of electricity to drive the pump. Obviously, a more advanced electric circuit is necessary to optimize the voltage and current of the generator and pump. From this experimental result, the best power density of the present packed µ-ORC generator was 214 W/m3 (214 mW/L) based on the volume of the packing box, 293 mW/kg based on the total mass excluding the heat and cold source fluid inside the heat exchangers, and 33.3 W/m2 (3.33 mW/cm2) based on the footprint (area of base) of the packing box. If we can double the expander efficiency and halve the packed size, the power density would be improved to approximately 5.5 times higher than the current value.

4. Conclusion In this study, we demonstrated a compactly packaged µ-ORC generator that can generate electricity without an external power supply. Although the demonstrated net power of the µ-ORC generator was very low, the potential of a very small scale ORC system for power generation from highly distributed low-grade heat (