A method and a system for dynamic switching of thermoelectric generators

Abstract

A system for dynamically switching thermoelectric generators, comprising: at least three interconnected thermoelectric generators (A, B, C), each having a first terminal and a second terminal; a supply voltage terminal (VIN) and a ground terminal (GND); for each thermoelectric generator (A, B, C), a first dedicated switch (P) configured to connect the first terminal of the thermoelectric generator to the supply voltage terminal (VIN), a second dedicated switch (N) configured to connect the second terminal of the thermoelectric generator to the ground terminal (GND), and a third dedicated switch (S) configured to connect the second terminal of the thermoelectric generator to the first terminal of an adjacent thermoelectric generator; and a controller configured to control the states of the dedicated switches so as to dynamically reconfigure the thermoelectric generators.

Description

[0001] A METHOD AND A SYSTEM FOR DYNAMIC SWITCHING OF THERMOELECTRIC GENERATORS

[0002] TECHNICAL FIELD

[0003] The present invention relates to a method and system for dynamically reconfiguring thermoelectric generators (TEGs) to exploit uneven temperature distributions in wearable devices and Industrial Internet of Things (loT) applications. More specifically, it focuses on selecting the optimal interconnection of TEG arrays in real time to maximize energy conversion efficiency under changing thermal conditions.

[0004] BACKGROUND

[0005] Thermoelectric generators (TEGs) are devices that exploit the Seebeck effect to convert temperature differentials into electrical energy. In a circuit made of two materials with different Seebeck coefficients (AS 0) connected electrically in series and thermally in parallel to a heat source and a heat sink, a Seebeck voltage Vsarises according to the relationship:

[0006] Vs= AS ■ AT, where AT is the difference between the hot side temperature (Th) and the cold side temperature (Tc). When a thermoelectric generator with internal resistance Ri is placed in series with a resistive load RL , the resulting current is I = Vs / (RL + Ri) and the generated electrical power P is:

[0007] In practice, TEGs can be deployed on surfaces having non-uniform temperature distributions, thereby harvesting energy from ambient or waste heat. However, in conventional designs, the arrangement of TEGs (and the electrical connections between them) is fixed at the time of manufacture based on an expected thermal distribution. In wearable devices or in Industrial Internet of Things (loT) systems, where ambient conditions and heat-generation patterns can fluctuate significantly, such static layouts often fail to achieve optimal power output once the temperature distribution deviates from initial assumptions.

[0008] WO2019043482A1 discloses a wrist- worn device with an array of TEGs that can be selectively activated or deactivated to heat or cool specific regions of the user’s wrist. Although this approach enables localized thermal management, it does not address optimizing the arrangement of TEGs for improved energy generation when temperature distributions vary over time.

[0009] SUMMARY OF THE INVENTION

[0010] Therefore, there is a need for a system and method that can dynamically reconfigure the interconnections among thermoelectric generators based on the real-time thermal environment, particularly to maximize the overall energy conversion efficiency.

[0011] The inventor has determined that the total power output of multiple TEGs can be significantly enhanced if their connections are selectively and dynamically switched to match the changing temperature gradients present across each individual generator. By allowing realtime reconfiguration (for instance, choosing between series or parallel groupings, or selectively disabling certain TEGs), the overall internal resistance can be tuned to more closely match the load. This dynamic matching maximizes power extraction under conditions that may vary in ways that were unpredictable at the device’s design stage.

[0012] The method and system according to the invention involve specific hardware components and operational steps for physically altering the electrical pathways among thermoelectric generators, resulting in tangible changes in the flow of current and voltage distribution. This concrete implementation is rooted in a particular physical arrangement of switches, controlled by a microcontroller or similar device, that exerts direct technical influence on the power output of the TEG array. Consequently, the invention addresses a technical problem: suboptimal utilization of temperature gradients and provides a technical solution that optimizes the TEG interconnections in real time, producing verifiable gains in electrical energy harvesting.

[0013] Furthermore, the dynamic reconfiguration of TEGs yields a non-trivial technical effect: the system actively adjusts each generator’s contribution by altering switch states so as to better match the load and temperature gradient variations. This involves integrated hardware elements (e.g., transistors, microcontrollers) operating cooperatively to generate and harness real electrical signals.

[0014] As demonstrated herein, when operating with low temperature gradients (on the order of 1-2°C) that often arise in wearable devices and certain industrial environments, dynamic reconfiguration can yield substantial gains in output power compared to static TEG arrangements. This approach is particularly beneficial for wearable applications and loT devices powered by energy-harvesting systems mounted on machinery with fluctuating heatgeneration profiles. By allowing the system to adapt in real time, the invention offers improved thermal-to-electrical conversion efficiency and provides a practical route to more effective energy harvesting using TEG arrays in scenarios where temperature gradients are small and / or prone to change.

[0015] In a specific aspect, the invention relates to a system for dynamically switching thermoelectric generators, comprising: at least three interconnected thermoelectric generators, each having a first terminal and a second terminal; a supply voltage terminal and a ground terminal; for each thermoelectric generator, a first dedicated switch configured to connect the first terminal of the thermoelectric generator to the supply voltage terminal, a second dedicated switch configured to connect the second terminal of the thermoelectric generator to the ground terminal, and a third dedicated switch configured to connect the second terminal of the thermoelectric generator to the first terminal of an adjacent thermoelectric generator; and a controller configured to control the states of the dedicated switches so as to dynamically reconfigure the thermoelectric generators.

[0016] In another aspect, the invention relates to a method for dynamically switching thermoelectric generators in a system as described herein, the method comprising: measuring the open-circuit voltage of each thermoelectric generator; generating a set of all valid unique circuit configurations for the thermoelectric generators; computing the maximum output power for each configuration within the set; identifying the configuration that provides the highest maximum output power; and setting the states of the first, second, and third dedicated switches for each thermoelectric generator according to the identified configuration.

[0017] In a preferred embodiment, the controller in the system is further configured to selectively activate or deactivate each thermoelectric generator by opening or closing its dedicated switches, optimizing energy conversion efficiency based on the current thermal environment. It allows individual TEGs to be switched off when their contribution to total power is minimal or detrimental, thereby conserving power and improving overall performance.

[0018] In another preferred embodiment, the dedicated switches for each thermoelectric generator are metal-oxide-semiconductor field-effect transistors (MOSFETs). MOSFET-based switching can be achieved with low on-resistance and low drive power, enabling efficient and reliable control over the TEG array connections.

[0019] In a further preferred embodiment, the controller adjusts the switches so that the internal resistance of the thermoelectric generators collectively aligns with the load resistance. It ensures the system operates near the optimal load matching condition, thereby enhancing power extraction under variable load and thermal scenarios.

[0020] In a still further preferred embodiment, the foregoing method is repeated at predetermined intervals, allowing the system to adapt to changes in temperature gradients over time. This allows the continuous tracking of environmental conditions, ensuring the TEG arrangement remains at or near peak power output without manual intervention.

[0021] In another preferred embodiment, the generation of valid circuit configurations is accomplished by interpreting each configuration as a binary sequence that indicates whether each thermoelectric generator’s second terminal connects to an adjacent generator or to ground, along with whether its first terminal connects to the supply voltage. It restricts the possible circuits to a tractable set (2n-n), significantly reducing computational complexity for real-time implementation.

[0022] In a further preferred embodiment, the chosen configuration ensures that the collective internal resistance of the TEGs matches the load, resulting in maximal power transfer. It allows the system to fine-tune the circuit to achieve the highest efficiency possible, even when operating under low temperature gradients or in rapidly changing thermal environments.

[0023] BRIEF DESCRIPTION OF DRAWINGS

[0024] The present invention is shown by means of example embodiments on drawings, wherein:

[0025] Fig. 1 A shows a general schematic of the electrical configuration of the system according to the invention.

[0026] Fig. IB presents a more detailed schematic of an exemplary system implementation.

[0027] Fig. 2 contains a table illustrating possible circuit configurations for various numbers of thermoelectric generators.

[0028] Fig. 3 depicts how a series circuit can be transformed into an equivalent circuit comprising an ideal current source and an internal resistance.

[0029] Fig. 4 shows the circuit after simplification of all series connections.

[0030] Fig. 5 is a graph of output power as a function of load resistance for multiple circuit configurations.

[0031] Figs. 6 and 7 compare power outputs for selected configurations under different temperature gradients.

[0032] Figs. 8 and 9 illustrate changes in the shape of the output power curves when temperature differentials vary among multiple generators.

[0033] Fig. 10 summarizes the improvement in output power obtained by regrouping thermoelectric generator connections.

[0034] Fig. 11 uses violin plots to show how output power improves as the range of temperature gradients widens for four-generator and eight-generator systems. Fig. 12 provides an algorithm (in pseudocode) for generating all valid circuit configurations.

[0035] Fig. 13 provides an algorithm (in pseudocode) illustrating how switch states are set to implement each chosen circuit configuration.

[0036] DETAILED DESCRIPTION OF EMBODIMENTS

[0037] 1. Overview of the system

[0038] When multiple thermoelectric generators (TEGs) are installed between a flat heat sink and a flat heat source (both having perfectly aligned surfaces and uniform heat distribution) each TEG becomes indistinguishable from an electrical standpoint (assuming each generator has the same parameters, such as internal resistance). Under these ideal conditions, the maximum output power for a circuit of n TEGs is simply nP, where P is the maximum output power of a single generator. Typically, the only circuit-related factor that changes among different interconnections (e.g., series versus parallel) is the overall equivalent series resistance, which in turn influences the load resistance RL necessary to achieve maximum power.

[0039] In real-world energy-harvesting applications, however, TEGs seldom operate under identical temperature gradients. Instead, they are placed in different physical locations, e.g. on a person, animal, or machine and are exposed to slightly varying thermal conditions. As a result, each TEG experiences a unique temperature difference and thus a unique Seebeck voltage. While such variations may be less consequential at large temperature gradients, they have a substantial impact in low-gradient scenarios (e.g., a few degrees Celsius), where overall performance can be severely reduced if all TEGs are hard-wired in a single, unchanging configuration.

[0040] The inventor has observed that significant power gains can be achieved when the TEG interconnection scheme is allowed to adapt to these changing conditions, rather than remaining in a fixed arrangement. However, because the potential number of ways to interconnect n generators grows factorially with n, creating a fully flexible routing system quickly becomes impractical for larger arrays. To address this, the inventor has designed a circuit architecture in which the number of switching elements is capped at 3n. Even with this constraint, the system can still provide substantial performance benefits compared to an unchanging (static) configuration.

[0041] Figure 1A illustrates the conceptual schematic for an embodiment of the invention, limited to 3n switching elements. Each thermoelectric generator t is governed by three switches: - tP: connects the generator’s positive terminal to the supply voltage (VIN). - tN: connects the generator’s negative terminal to ground (GND).

[0042] - tS: Connects the generator’s negative terminal to the positive terminal of the next generator in the loop (using modulo n addressing).

[0043] For example, in a setup with three generators A, B and C, if the switches AP, AS, BS and CN are closed while all others remain open, then A, B and C form a series circuit. Although Figure 1A focuses on n=3, the same approach extends to any number of thermoelectric generators.

[0044] Dynamic switching can be handled by a low-power microcontroller (such as an STM32L151), and the switches themselves can be implemented using transistors - for instance, N-channel MOSFETs such as A03400 devices. An example of this practical implementation is shown in Figure IB.

[0045] Figure 2 provides a table showing all possible connection configurations for n ranging from 1 to 4, but it also helps illustrate how similar configurations can be derived for larger n. In that table, each grouping in parentheses denotes which TEGs are in series. Observing that these groupings correspond to partitions of a loop, it can be shown that the number of possible loops is (2n- n) which is significantly smaller than the total number of all imaginable interconnections. Even so, this reduced set of configurations still enables meaningful improvements in power output relative to a static design.

[0046] In the arrangement of Figure 1, it is assumed that each TEG maintains the same polarity and remains connected (i.e., it is neither shorted out nor left open-circuited). Within these constraints, the system can form parallel groupings of one or more sets of series-connected generators, or it can disconnect a particular TEG if desired.

[0047] Each TEG can be modeled as a non-ideal voltage source, represented by an ideal voltage source Ei in series with an internal resistance Rint. In some scenarios, certain generators might be grouped in series, with these groups then connected in parallel between VIN and GND.

[0048] According to Thevenin’s theorem, any series cluster of generators can be simplified to a single voltage source Eeq= En +... + Eikwith internal resistance Req= k Rint.

[0049] By Norton’s theorem, each such series cluster becomes an equivalent circuit consisting of an ideal current source Jeq= Eeq / Reqin parallel with the internal resistance Req.

[0050] Figure 3 depicts how a series circuit is transformed into a current source plus an internal resistance, and Figure 4 shows a circuit after all its series connections have been simplified.

[0051] It can then be further transformed into a circuit consisting of a single current source Jeqby summing the individual current sources from each parallel branch, (Jeq= Jji +... + Jji) and calculating the overall equivalent resistance as:

[0052] Since the objective is to identify the configuration that yields the highest output power among all possible interconnections, it is often practical to assume (for ease of calculation) that each TEG’s internal resistance is a uniform value (e.g., normalized to unity). This assumption streamlines comparisons across multiple configurations.

[0053] 2. First confirmation of advantages.

[0054] This first example focuses on the system illustrated in Figure 1. Within this system, each of the nine switches can be independently opened or closed, giving 29= 512 possible switch configurations. However, not all of these combinations form a valid electrical circuit. For example, if switches AP, BP, CP, AN, BN and CN are open, none of the generators would connect to VIN or GND. Consequently, only 458 of the 512 potential configurations produce a properly connected circuit.

[0055] Each configuration c E C = (AP, BP, CP, AS, BS, CS, AN, BN, CN) can be uniquely associated with a vector v = (vi,..., V9) G [0, I]9, where each component corresponds to one of the switches in the set. A value of 1 in the vector indicates that the switch is closed, while 0 indicates it is open.

[0056] Apart from confirming which configurations are valid, the behavior of configurations over a range of load resistances RL can be examined. Figure 5 illustrates the output power as a function of RL obtained for all 512 configurations, calculated via LTspice, under the following assumed generator parameters: AS = 36,87 mV / K, Ri = 2,76 Q, K = 333 mW / K.

[0057] The following temperatures were assumed: cold side temperature Tc= 34°C, hot side temperatures Thi= 36°C, Th2= 35°C; and ambient temperature Ta= 25°C.

[0058] As shown in Figure 5, only a few configurations reach the maximum power point.

[0059] Based on these findings, the inventor concluded that only five unique configurations - the ones listed in the table of Figure 2 - truly need to be evaluated. Other possibilities are either isomorphic (electrically equivalent) or do not utilize all the generators. For the case where all three generators experience the same temperature gradient, the best performance occurs in pure series or parallel connections, as shown in Figure 6 (configurations numbered 455, 334, 213, 92, and 419, with some graphs overlapping).

[0060] However, if two generators experience a temperature gradient increase of 1°C while the third remains at the original gradient, other configurations can now achieve maximum power (Figure 7). This shift in the relative shape of the power curves is further demonstrated in Figures 8 and 9, where two generators have a constant temperature gradient and the third’s gradient varies around 6°C.

[0061] As the variation in temperature gradients among the generators grows, the benefit of reorganizing their connections also increases. Figure 10 provides a summary of the performance improvements when a 3°C difference is introduced, underscoring the advantages of dynamic reconfiguration.

[0062] 3. Second confirmation of advantages.

[0063] In this second example, the analysis from the first example is extended to cover a broader range of temperature gradient variations. To begin, the cold-side temperature Tc= 34°C, the ambient temperature Ta= 25°C, and the initial hot-side temperature for every thermoelectric generator (TEG) Th = 36°C. From this baseline, each generator’s hot-side temperature is randomly varied according to a uniform distribution within the range [-1.8, +1.8] relative to 36°C.

[0064] After these variations are applied, the standard deviation of the hot-side temperatures across all generators is calculated for each simulated circuit configuration. Figure 11 presents the results using violin plots. Similar to box plots, violin plots display markers for medians and boxes for interquartile ranges, while also showing the probability density of the data across different values.

[0065] The left side of Figure 11 pertains to systems with four generators, whereas the right side pertains to systems with eight generators. In both cases, there is a clear upward trend in maximum output power (y-axis) as the standard deviation of the hot-side temperatures (x-axis) increases. This demonstrates that greater disparity in individual TEG temperature gradients amplifies the performance benefits of dynamic reconfiguration.

[0066] 4. Example implementation of the routing algorithm

[0067] In this example, the routing procedure proceeds as follows:

[0068] - Measure open-circuit voltage (VOC) of each generator: one by one, each generator is isolated from the circuit, while the others remain disconnected. The open-circuit voltage VOC of the selected generator is then measured.

[0069] Generate all possible unique circuit configurations: a sequence of binary numbers representing all valid connection schemes is created. This generation process ensures each combination corresponds to a legitimate way of wiring the generators (i.e., no invalid or redundant circuits). - Evaluate and identify maximum output power: for each unique circuit configuration, the system’s output power is computed. Once the power calculations are complete, the configuration that yields the highest power is identified.

[0070] - Reconfigure the circuit: the controller then sends the corresponding switch states to the General-Purpose Input / Output (GPIO) pins, which control the transistors (switches) and implement the optimal circuit configuration.

[0071] This process is repeated at set intervals, enabling real-time adaptation to changing temperature distributions and other environmental factors.

[0072] Figure 12 presents the algorithm in pseudocode for generating all the valid sequences, which can be adapted for specific microcontrollers by modifying the data types or system calls. In this algorithm, each thermoelectric generator i can optionally connect to generator (i + 1 mod n), wherein n is the number of generators in the sequence. That is, for n generators there are 2nways of connecting them (or not) to successive generators. Thus, there are (2n- 1) possible ways to form loops within n generators (excluding the single configuration that connects all generators in series).

[0073] For each subsequence within the overall sequence, the first generator of that subsequence is connected to the VIN terminal, and the last generator is connected to the GND terminal. Any subsequences that include a total of (n-1) connections are considered electrically equivalent; therefore, only one such subsequence is retained while the remaining (n-1) subsequences of this type are excluded. As a result, there are (2n- n) circuits that are electrically distinct.

[0074] In particular, line 1 of the algorithm initializes the set that will store valid subsequences. Then, in line 3, the algorithm iterates over all possible sequences from zero to (2n- 1) (excluding the single sequence in which all generators are connected). In line 4, the number of connections in each sequence is determined by counting the ones in its binary representation. Should the algorithm encounter a sequence with exactly (n-1) ones that has not yet been added, a flag is set to prevent the addition of any other sequence having the same number of connections. Finally, in line 10, the newly found sequence is added to the result set.

[0075] The binary sequences generated by the algorithm shown in Figure 12 can be translated directly into switch states that connect each generator to the VIN, GND, and adjacent generator terminals in the sequence. The specific algorithm for setting these switch states in a circuit with n generators is illustrated in Figure 13. It treats each input sequence as a binary number of length n and iterates through all of its bits (line 1). Initially, all switches connecting the i-th generator to VIN, GND, and the next generator are set to the open (non-conducting) state. If the i-th bit in the binary representation is zero, the switch connecting the i-th generator to GND is closed (conducting). The switch connecting the i-th generator to VIN is then closed only if the (i+1) -th bit (modulo n) is one and the i-th bit is zero (line 6). Likewise, the switch connecting the i-th generator to the next generator remains open if the i-th bit is zero and closes when that bit is one (line 8). Since the states of the switches leading to VIN and GND depend in part on how neighboring generators are connected, the sequence numbers used in these algorithms end up being three times shorter than what would be needed to represent every individual switch state. This compression reduces the computational workload and accelerates the calculation process.

Claims

CLAIMS1. A system for dynamically switching thermoelectric generators, comprising: at least three interconnected thermoelectric generators (A, B, C), each having a first terminal and a second terminal; a supply voltage terminal (VIN) and a ground terminal (GND); for each thermoelectric generator (A, B, C), a first dedicated switch (P) configured to connect the first terminal of the thermoelectric generator to the supply voltage terminal (VIN), a second dedicated switch (N) configured to connect the second terminal of the thermoelectric generator to the ground terminal (GND), and a third dedicated switch (S) configured to connect the second terminal of the thermoelectric generator to the first terminal of an adjacent thermoelectric generator; and a controller configured to control the states of the dedicated switches so as to dynamically reconfigure the thermoelectric generators.

2. The system according to claim 1, wherein the controller is further configured to selectively activate or deactivate each thermoelectric generator (A, B, C) by opening or closing the corresponding dedicated switches, thereby optimizing energy conversion efficiency based on current thermal conditions.

3. The system according to claim 1 or 2, wherein each of the first, second, and third switches is a metal-oxide-semiconductor field-effect transistor (MOSFET).

4. The system according to any one of claims 1 to 3, wherein the controller is configured to adjust the switches such that the overall internal resistance of the thermoelectric generators is matched to a load resistance (RL).

5. A method for dynamically switching thermoelectric generators in a system according to any one of claims 1 to 4, the method comprising: measuring the open-circuit voltage (VOC) of each thermoelectric generator (A, B, C); generating a set of all valid and unique circuit configurations for the thermoelectric generators (A, B, C); computing the maximum output power for each configuration within the set; identifying the configuration that provides the highest maximum output power; andsetting the states of the first, second, and third dedicated switches (P, N, S) for each thermoelectric generator (A, B, C) according to the identified configuration.

6. The method according to claim 5, wherein the steps of measuring, generating, computing, identifying, and setting are repeated at predetermined intervals to adapt to changes in temperature gradients.

7. The method according to claim 5 or 6, wherein generating the set of all valid circuit configurations is performed by interpreting each configuration as a binary sequence that indicates whether the second terminal of each generator is connected to an adjacent generator or to the ground terminal (GND), and / or whether its first terminal is connected to the supply voltage, thereby producing a reduced set of (2n-n) unique circuits for n thermoelectric generators.

8. The method according to any of claims 5 to 7, wherein the set configuration has the internal resistance of the system matched to the load.

Images