• Kurt Engelbrecht (Supervisor)

Activity: Examinations and supervisionSupervisor activities


The aim of this work is the experimental proof of concept of an energy harvester using shape memory effect and piezoelectricity to transform input thermal power in output electricity. The practice of recovering wasted or unused energy from the environment and making it available in form of electrical charge – i.e. energy harvesting – is attracting increasing attention since it can be employed for feeding small wireless autonomous devices as a valid alternative to batteries or grid connection. As a matter of facts, traditional solutions imply continuous maintenance (e.g. batteries need to be substituted periodically) or non-economical arrangements (e.g. connecting to the grid very small devices in remote locations is not only complicated but also anti economical). Energy harvesting then addresses the problem of feeding low-power electronics and autonomous wireless sensors (requiring few mW of electricity).
Since this field of research is pretty new, not much literature is available on the topic but on the other side lot of work and improvement can be done; moreover, there is not a recognized best way of doing things, then experimental activity needs to be performed to explore possible solutions.

In this work the problem of energy harvesting is addressed to considering thermal energy as available source input, to be transformed in electrical power; this is the starting point: all that comes after is the result of studies performed within this thesis. The first step is deciding which mechanism to be employed: the choice is to use shape memory effect of Ni-Ti for converting thermal power into mechanical one and then a piezoelectric actuator that receives mechanical power as an input and converts it into electricity. The reason for that is double: on one side, a research group already provided a similar device that could be used as reference and improved; on the other, Ni-Ti is the object of ongoing research due to its incredible properties and then it is due to contribute in experiments by providing another type of application for this material.
The second step is then gathering information on shape memory alloys and piezoelectric components from literature and manufacturers and designing an experimental session to provide the other pieces of information needed for the realization of this device.

Starting from shape memory effect, it is a property exhibited by certain materials that can restore the original shape of a plastically deformed sample by simply heating it as a consequence of a crystalline phase change – called martensitic transformation. In particular, at low temperature, below the transformation starting one, the material is in martensitic phase, which is soft and can be deformed quite easily. Then, when the specimen is heated up above a transition starting temperature, it recovers its original pre-deformation shape and converts the material to its high strength – austenitic – condition. The process is reversible, meaning that the same transformation occurs while cooling, even though some hysteresis can be detected and temperature range is slightly different. If the specimen is in a constrained configuration, i.e. it cannot recover its initial shape upon heating due to zero displacement constrains, force starts growing inside it; then, it goes down while cooling. These properties are known from literature. The problem is that it is not clear how force depends on temperature and which are relevant variables affecting performance: for this reason an experimental characterization is performed. First of all, a shape memory alloy is selected, in particular Ni-Ti due to its above the average properties, even though there are also other different classes of materials showing this effect. Then a bundle of wires of Ni-Ti is tested, considering that output force depends on cold wires temperature, hot wires temperature and pre-stress, meaning force applied to the specimen at cold state, before it is heated up and actuation, i.e. force, produced. Characterization is made for a limited but still meaningful temperature values and for a wide range of pre-stress. Results show that the best operating temperatures are 5 °C as cold temperature and 55 °C as hot one, being temperature difference fixed at 50 °C. Force difference between hot and cold states increases with pre-stress almost linearly; for this reason, a pre-stress of 1000 N is selected as the one to be employed in the device during operations. Actuation provided with these parameters is expected to be 600 N.

Moving to piezoelectricity, it is the property of producing electrical charge when submitted to a certain pressure (direct piezoelectric effect), manifested by a specific class of materials. Nowadays there are many devices employing the opposite effect – inverse piezoelectricity: they receive voltage as input and produce displacement as an output; knowing the relation between displacement and voltage it is possible to provide precise control. Among these materials, the most common one is Lead Zirconate Titanate due to its marked piezoelectric effect. Working principle is as follows: when subjected to a force, there is a shifting of electrons in the crystal structure of this material, resulting in a charge; this process is reversible in a certain range of temperature, force and electric field, meaning that if force is removed the electrons go back to initial configuration and charge goes back to zero. While inverse piezoelectric effect is well known, direct one is pretty unstudied due to limited number of practical applications: for this reason a sample of Lead Zirconate Titanate needs to be tested and characterized. The goal is to find how voltage depends on force (current is expected to be very small, while voltage relevant, for this reason it is selected as variable for evaluating the performance). In the experimental session the piezoelectric actuator is then compressed at increasingly high force and voltage is measured; moreover, the effect of compression speed is investigated since it could be a relevant variable considering that a transformation in the crystal occurs. Results show that absolute value of voltage increases almost linearly with force: when compressed the actuator produces positive voltage; then if it is shortened at maximum force and gradually released to zero stress, a negative voltage builds up: for this reason it is convenient to speak about potential difference per cycle. At 600 N, which is the force provided by Ni-Ti, potential difference is expected to be between 25 V and 35 V in relation to compression speed, which slightly affects output.

Once the properties of the two main components are clear, a device employing them is designed and built. A globally zero displacement assembly is chosen: Ni-Ti wires are clamped together with the piezoelectric actuator and then subjected to temperature fluctuation, so that they produce a force that alternatively compresses and releases the piezo, finally generating in this way an electrical charge. Water is selected as heat transfer fluid and 13 wires of 5 mm length and 0.8 mm diameter of active part are disposed circularly, perpendicularly to the flow in an optimized configuration in terms of heat transfer, held in place by plastic plates and put inside a cylindrical case. A shaft then puts the active part in connection with piezoelectric actuator, well separating it from the flow since it is not waterproof.
While designing this assembly, great care is reserved to heat transfer problem and to a detailed study of how stress/strain state should evolve during operation. As regards heat transfer, both manual computation and numerical simulations are performed in order to find a configuration where water and wires exchange heat properly, i.e. effectively, homogeneously and fast. As a matter of facts, non-homogeneous heat transfer with the wires would result in uneven shape memory effect and then in differential displacement; moreover, being the slowest phenomenon, heat transfer affects operating frequency of the whole device: for this reason it needs to be analyzed carefully. Stress/strain problem is also studied in details, cross-referencing data from both literature and experiments: it is fundamental to design the length and number of wires compatibly with stress and displacement required by piezo, which in turn depends on shape memory effect at a given temperature. In other words, it is mandatory to run a trial and error process, where there are different involved variables that need to be compatible one with the others in order to have the device working. These studies show that aforementioned configuration of 13 wires of 5 mm length and 0.8 mm diameter is a solution to this problem; even though it is not the optimized one, it should work properly for a proof of concept.

Getting back to design, there are then auxiliary elements, in particular a hydraulic circuit composed of two identical lines, one for the hot and one for the cold side, alternatively switched by means of three way valves. Water baths provide temperature control and water flow. Their goal is exchanging heat with active part of the device, providing desired temperature change at a given frequency. The other auxiliary element is the electrical circuit, aimed at collecting and measuring the charge produced by the piezo; the main components are a series of two resistors of known resistance coupled with a voltmeter for reading the voltage and a relay for switching the circuit on and off. In particular, since generated charge is very small, it is necessary first to build up a voltage across the piezo and then to discharge it across the measurement system within a certain frequency.

Finally, there is a control/data acquisition system based on LabVIEW aiming at measuring temperatures with thermocouples and recording it and at automatically switch electrical circuit with a small control current while recording voltage. Once all the parts are designed, they are built and assembled and tests are run on the apparatus. Experimental setup consists in the aforementioned components, clamped with a mechanical actuator and operating as follows. The mechanical actuator provides 1000 N of pre-stress while maintaining a zero displacement configuration; the assembly made of a cylinder containing Ni-Ti active component and a shaft for rigidly transmitting force to the piezo chip is blocked between the two plates of the mechanical actuator. Two hoses connect active chamber to hydraulic circuit while two wires connect piezo to electrical circuit. Data acquisition system completes the setup. As regards operation, several experiments are done to find best working point. Sequence of operations is as follows: pre-stress is provided while cold water is circulating, then circuit is closed to discharge voltage built up by the pre-stress and opened after that; these are preliminary actions. Then, three-way valves are switched so hot water flows and heats up the wires, which expand and actuate on the piezo, generating a voltage that can be measured after closing the circuit; the circuit is then re-opened, valves switched to cold water and then a negative voltage builds up across the piezo; electrical circuit is closed again and negative voltage measured, then it is re-opened. At this point, a new cycle begins. Data acquisition saves temperature to time and voltage to time.

These data are then elaborated employing a MATLAB script where total voltage, current, power and energy harvested per cycle are measured. Results are very positive: a potential difference of 30.73 V and a current of 2.82 mA are detected; power is 44.46 mW, while harvested energy per cycle is 0.33 mJ. An efficiency index which reports harvested energy per cycle to the mass of active part – Ni-Ti – is defined and computed: it is 2.04 mJ/g. These numbers not only prove the concept of energy harvesting, but also highlight a performance that is much better than previous devices and then can be considered as a good starting point for future developments.

In particular, an optimization process of existing device is suggested as it could improve the performance dramatically. In the following there is a list of suggestions on what could be done. First of all, an observation is mandatory: this thesis is a proof of concept, meaning that there are no requirements in terms of performance, dimensions and output of the device; when designing an energy harvester for a specific application, starting point should always be electrical requirements from the load and then design should follow up. The point is that completely different devices could be built according to the kind of application: the only element in common would be the working principle. Future developments for this specific design can be summarized as follows. First of all, from a Ni-Ti point of view it could be interesting to increase the number of wires for providing a higher actuation and a bigger pre-stress as operating condition, impossible now due to limited yield strength as failure tests show. Another improvement could be made in relation to hydraulic circuit and heat transfer: in this device operating frequency is fairly slow because the temperature of heat transfer fluid is the same as Ni-Ti desired one and valves are manual. It could be possible to design automatic valves and employ higher temperature difference so that desired temperature of Ni-Ti would be reached faster and at this point valves would switch automatically. As a result, system would operate with higher frequency and then also global harvested energy would be bigger, due to more cycles in the same time interval. Finally, a wider characterization of temperature influence on Ni-Ti displacement could lead to the definition of better working conditions.
Period2 Oct 20162 Apr 2017
ExamineeLuigi Bazzan, Luigi Bazzan, Luigi Bazzan, Luigi Bazzan, Luigi Bazzan, Luigi Bazzan, Luigi Bazzan, Luigi Bazzan, Luigi Bazzan, Luigi Bazzan, Luigi Bazzan & Luigi Bazzan
Examination held at
  • Polytechnic University of Milan
Degree of RecognitionInternational


  • Elastocaloric
  • Energy harvesting