- Email: [email protected]
Sensor-less control of the methanol concentration of direct methanol fuel cells at varying ambient temperatures
Applied Energy 129 (2014) 104–111
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Sensor-less control of the methanol concentration of direct methanol fuel cells at varying ambient temperatures Myung-Gi An, Asad Mehmood, Heung Yong Ha ⇑ Energy Convergence Research Center, Korea Institute of Science and Technology (KIST), Seoul 136-791, Republic of Korea Department of Energy and Environmental Engineering, Korea University of Science and Technology (UST), Yuseong-gu, Daejeon 305-333, Republic of Korea
h i g h l i g h t s A new algorithm is proposed for the sensor-less control of methanol concentration. Two different strategies are used depending on the ambient temperatures. Energy efficiency of the DMFC system has been improved by using the new algorithm.
a r t i c l e
i n f o
Article history: Received 2 February 2014 Received in revised form 24 April 2014 Accepted 28 April 2014
Keywords: Direct methanol fuel cell Sensor-less control Methanol concentration Stack temperature Ambient temperature
a b s t r a c t A new version of an algorithm is used to control the methanol concentration in the feed of DMFC systems without using methanol sensors under varying ambient temperatures. The methanol concentration is controlled indirectly by controlling the temperature of the DMFC stack, which correlates well with the methanol concentration. Depending on the ambient temperature relative to a preset reference temperature, two different strategies are used to control the stack temperature: either reducing the cooling rate of the methanol solution passing through an anode-side heat exchanger; or, lowering the pumping rate of the pure methanol to the depleted feed solution. The feasibility of the algorithm is evaluated using a DMFC system that consists of a 200 W stack and the balance of plant (BOP). The DMFC system includes a sensor-less methanol controller that is operated using a LabView system as the central processing unit. The algorithm is experimentally confirmed to precisely control the methanol concentration and the stack temperature at target values under an environment of varying ambient temperatures. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Direct methanol fuel cells (DMFCs) continue to garner interest as a clean energy technology for powering small and portable electronic devices because of positive characteristics such as high-energy density, flexibility in power output from sub-watt to several hundred watts, and reliable operation. Over the past decade, a number of worldwide efforts have been focused on improving the performance and durability of DMFCs [1–9]. However, DMFCs still suffer from low efficiency because of poor electrode kinetics and methanol crossover from the anode to the cathode through the polymer electrolyte membrane [10,11]. The methanol crossover leads to not only fuel waste, but also mixed potential at the cathode, which adversely affects the performance and fuel ⇑ Corresponding author at: Energy Convergence Research Center, Korea Institute of Science and Technology (KIST), Seoul 136-791, Republic of Korea. Tel.: +82 2 958 5275; fax: +82 2 958 5229. E-mail address: [email protected] (H.Y. Ha). http://dx.doi.org/10.1016/j.apenergy.2014.04.100 0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.
efficiency [12]. The methanol crossover rate is mainly influenced by the concentration of the methanol feed at the anode and by the operating temperature of the DMFC [13]. Controlling the methanol concentration in an adequate range therefore reduces the methanol crossover under given operating conditions, and plays a critical role in stable and efficient DMFC operations. In order to achieve this goal, electronic sensors are generally installed in a feed re-circulating DMFC system to monitor and control the methanol concentration. However, the sensors are known to have many problems in terms of cost, size, durability and reliability. Zhao et al. [14] reviewed various methanol sensors for DMFCs that are generally classified into two groups: electrochemical and physical. Physical sensors are reliable and have a wide measurement range, but they are expensive and too bulky for use in portable DMFC systems. Electrochemical sensors are known to have a narrow sensing range and a slow response time due to diffusion limitations, which hampers their reliability. These also have durability issues due to a degradation of the electrolyte membranes and a deterioration of the electrodes with operating time. Some efforts have been made
M.-G. An et al. / Applied Energy 129 (2014) 104–111
to solve the problems of electronic sensors, and, as a result, sensorless methanol controllers have been proposed and reported by some research groups. Several researchers have reported sensor-less methanol control using consumption equations rather than sensors to estimate the methanol concentration in the feed [13,15,16]. Chiu and Lien [13] proposed a three-dimensional measurement space (CCS) and an interpolation algorithm (ICCS) based on the constant-concentration surfaces. Once the CCS (I, V, T) is obtained for the known methanol concentrations in the feed, the unknown methanol concentration that corresponds to the in situ-measured data (Iu, Vu, Tu) can be estimated by interpolation. By using the ICCS, they estimated the methanol concentrations in the feed based on the current, voltage, and cell temperature of a fuel cell. Ha et al. [15] devised a sensor-less algorithm that could control the methanol concentration at a set value by supplying the same amount of methanol consumed in a DMFC system to the recirculating methanol feed solution. They built a database of methanol consumption rates that were collected under various operating conditions and utilized the database to calculate the amount of methanol needed to maintain the methanol feed concentration at a set value. Shen et al. [16] reported a real-time fuel control algorithm based on Chiu and Lien’s ICCS (I, V, T) algorithm. They modified the ICCS algorithm by accounting for the ‘‘MEA decay’’ and including an in situ estimating method in their control program that could estimate the methanol and water consumption quantities by accumulating the operating time and the methanol and water consumption rates. The program could be used to determine the remaining amount of methanol in the operating DMFC system that could be used to control the methanol concentration by compensating for the depleted methanol feed solution. On the other hand, other studies [17–23] have reported sensor-less methods that utilized the DMFC operating characteristics as feedback parameters to control the methanol concentration in the feed. Chang et al. [17,18] reported a sensor-less algorithm that was referred to as ‘‘impulse response based on discrete-time fuel-injection’’ (IR-DTFI) along with a modified version that could regulate the fuel concentration in order to optimize fuel cell performance by accounting for the changes in the characteristic values of a DMFC stack, such as voltage, current, and power, during a set period of operation. They [19] also suggested an advanced version referred to as ‘‘impulse response based on current-integral and discrete-time fuel-injection’’ (IR-CIDTFI) algorithm that could shorten the monitoring period (5 s or less) by calculating the amount of fuel consumed during the last monitoring cycle for faster system response and greater stability. They tested the performance of IR-DTFI with a 40 W DMFC system to power portable electronics and to evaluate the effectiveness of both IRDTFI and IR-CIDTFI algorithms from the point of operating characteristics [20,21]. Arisetty et al. [22] found that maximum voltage could be obtained by adjusting the methanol concentration in the feed under a given current density. Based on this finding, they developed an in situ sensor-less methodology that employed the cell voltage as the feedback to optimize the methanol concentration for maximum power density under dynamic operating conditions while maintaining a high level of fuel utilization. Lian and Yang [23] developed a sensor-less adaptive fuel concentration control (SAFCC) algorithm to regulate the methanol concentration in a suitable range by detecting transient voltage behavior under pulse-like changes in the load current. Although those sensor-less controllers are believed to be useful in controlling the methanol concentration, no controller has ever accounted for the effect of the ambient temperature surrounding a DMFC system. Our first version of the sensor-less methanol concentration controller [15] was designed to regulate the methanol concentration in the feed by supplying an amount of pure methanol to the recirculating methanol feed solution that would be equal to the amount of
105
methanol consumed in a DMFC system. Later, the algorithm was revised to improve its accuracy by using the stack temperature as a feedback parameter, which has a close correlation with the methanol concentration [24]. This algorithm has been further modified during this study by including the ambient temperature as a feedback parameter. Ambient temperature affects the stack temperature to allow deviation from a set value. Change in the stack temperature can be recovered by adjusting either the cooling rate of the anode heat exchanger or the pumping rate of the pure methanol to the depleted feed solution. When the ambient temperature is lower than the reference temperature (i.e., 23 °C in this study), the stack temperature decreases to below the set value because of excessive heat loss from the stack. In this case, the temperature of the circulating methanol solution can be raised by decreasing the cooling rate of the heat exchanger to compensate for the heat loss from the DMFC stack while the pumping rate of the pure methanol remains unchanged. On the other hand, when the ambient temperature is higher than the reference value, however, the heat loss from the stack is lowered and therefore the stack temperature increases above the set value. In this case, the pure methanol pumping rate can be lowered to decrease the methanol concentration in the feed, thus returning the stack temperature to the set value. The lowered methanol concentration reduces the methanol crossover rate, decreasing the heat generation at the cathode and lowering the stack temperature. In the present study, a new version of a sensor-less methanol concentration control algorithm has been proposed based on the feedback from ambient temperatures (SLCCFA). The effects of ambient temperature on the temperature and the feed concentration of a DMFC stack are explored in detail under various operating conditions, and these are used in designing a sensor-less algorithm. The feasibility of the new sensor-less algorithm has been evaluated in terms of the methanol concentration, stack temperature and system efficiency of a 200 WDMFC system.
2. Experimental setup 2.1. Building a database of methanol consumption rates in a DMFC stack In this control algorithm, the methanol consumption rates measured with a typical DMFC were used to calculate the amounts of pure methanol needed to compensate for the depleted feed solution under various operating conditions. The amounts of pure methanol to be pumped were determined by the experimental equations made in our earlier work [15]. The methanol consumption rates had been measured using a large-size single-cell DMFC with an active area of 150 cm2 under various operating conditions by changing the output current, methanol concentration in the feed (1.3, 2.6, 3.2, and 3.8 wt.%), and cell temperature (40, 60, and 80 °C) under fixed flow rates of air (1118 ml min1) and methanol feed (8.78 ml min1) based on a 3/3 stoichiometry (O2/ methanol). The total methanol consumption rate (Nm,t) in a DMFC is the sum of the methanol consumption rates by electrochemical oxidation (Nm,e) and methanol crossover (Nm,x) (Eq. (1)) [15].
Nm;t ¼ Nm;e þ Nm;x
ð1Þ
Nm,e was estimated based on the output current, and Nm,x was obtained by measuring the amount of CO2 at the cathode outlet that was produced by the oxidation of crossed-over methanol at the cathode. The amount of unreacted methanol that was exhausted from the cathode outlet was considered negligible because most of the crossed-over methanol was oxidized to CO2 at the cathode. However, a significant amount of CO2 generated
106
M.-G. An et al. / Applied Energy 129 (2014) 104–111
by the electrochemical oxidation of methanol at the anode penetrated through the membrane into the cathode. Therefore, Nm,x was estimated by the total CO2 flux measured at the cathode outlet ðN mCO2 Þ minus the CO2 crossover ðN xCO2 Þ (Eq. (2)). To measure the crossed-over CO2 from the anode to the cathode, a half-cell test was carried out using a fixed flow rate of nitrogen that equaled that of the air used for full-cell tests.
Nm;x ¼ NmCO2 NxCO2
catalyst on the anode and a 2 mg-Pt cm2 catalyst on the cathode. Graphite plates with a parallel serpentine-type flow field were used as separators. The flow rates of the methanol solution supplied to the anode, and the dry air supplied to the cathode, were set at 240 ml min1 and 23 L min1, respectively. The DMFC stack was operated on a constant electric load of 25 A (166 mA cm2) at a temperature of approximately 67 °C.
ð2Þ
The Nm,t was then calculated for the DMFC. In order to calculate the Nm,t for the 20-cell DMFC stack, it was multiplied by 20, and used for the operation of a DMFC stack under a constant electric load of 25 A. 2.2. Configuration of a sensor-less DMFC system The DMFC system was comprised of a methanol mixing chamber, three liquid pumps, a 200 W stack, two heat exchangers with air cooling fans, a pure methanol reservoir, a water reservoir, a liquid-level sensor, several thermocouples (K-type), and an air blower to the cathode, as shown in Fig. 1. The DMFC system was also equipped with a SLCCFA processor and a DC–DC converter. The SLCCFA processor simultaneously functioned as a central processing unit that regulated the operations of all the components in the DMFC system. Three liquid pumps were used: The first (a puremethanol pump) for pumping pure methanol into the methanol mixing chamber, the second (a main pump) for circulating a methanol feed solution along the recirculation loop passing through the DMFC stack, and the third (a water pump) for supplying condensed water from the cathode heat exchanger to the methanol mixing chamber. When the level of the methanol solution in the methanol-mixing chamber fell below the preset height, the water-feed pump started supplying water from the water reservoir to the mixing chamber. There were three thermo-couples (K-type) attached to the inlet (Tan.in) and outlet (T) of the anode side of the stack and the outlet (Th.out) of the anode heat exchanger to measure the temperatures of the methanol feed circulating along the fuel loop. The anode outlet temperature (T) was considered to be the temperature of the stack. In order to measure the methanol concentration, a methanol sensor (FC 10, ISSYS) was installed between the mixing chamber and the stack, which was used only for monitoring but not for controlling the methanol concentration. Each air-cooled heat exchanger made of stainless steel was integrated with a cooling fan. 2.3. DMFC stack specifications and test conditions The 200 W-class DMFC stack consisted of 20 single-cells with MEAs that had an active area of 150 cm2. The MEAs were purchased from Johnson-Matthey, Inc. and were made of Nafion 115 membranes and carbon paper electrodes with a 6 mg-PtRu cm2
3. Results and discussion 3.1. Behavior of the 200 W DMFC system at varying ambient temperatures First, background data were collected in order to design a sensor-less methanol concentration control algorithm. The experiments were carried out to evaluate the effects of ambient temperature on the behavior of the DMFC system. In these experiments, the DMFC system was operated at a steady state and the methanol concentration in the feed was controlled by supplying a fixed amount of pure methanol (2.1 g min1) to the recirculating methanol feed solution that was estimated to be consumed under given operating conditions of an electric load of 25 A, a stack temperature of 67 °C, a methanol concentration of 1 M (3.2 wt.%), and a stoichiometry (k) of both reactants at 3 (the flow rates of methanol solution and dry air were set at 240 ml min1 and 23 L min1, respectively) [15]. Fig. 2 shows the changes in the temperature, methanol feed concentration and output voltage of the DMFC stack as the ambient temperature (Tam) varies in the range of 12.5– 23.0 °C. The temperatures of the methanol solution were measured at the anode inlet (Tan.in) and anode outlet (T, considered equal to the stack temperature) of the stack and at the heat exchanger outlet (Th.out). In the beginning stage of the run from 0 to 40 min in Fig. 2a, the Tam is 23 °C and the stack temperature is maintained almost constant. When the Tam starts to decrease from 23.0 to 12.5 °C (at 40 min), the stack temperature decreases remarkably followed by drops in Th.out and Tan.in. The lowered stack temperature reduces the methanol consumption rate due to the reduced methanol crossover from the anode to the cathode and thus results in an increased methanol concentration in the feed from 3.2 to 3.4 wt.%, as shown in Fig. 2b. Afterwards, as the Tam increases again after 100 min, the stack temperature and the methanol concentration gradually return back to their initial values. In Fig. 2b, the output voltage also decreases corresponding to the decreased Tam even though the methanol concentration is increased above the target value in a time span of 60–120 min. This result shows that simply increasing the methanol feed concentration may not compensate for the decreased performance due to a lowered stack temperature. In this case, it may be better to seek a method that will deter a drop in the stack temperature. This could be achieved by lowering the cooling rate of the heat exchanger on the methanol circulation loop: the lowered cooling rate dissipates less heat from the
Fig. 1. Configuration of a sensor-less DMFC system with methanol re-circulation.
M.-G. An et al. / Applied Energy 129 (2014) 104–111
Fig. 2. Profiles of the changes in (a) the temperatures, and (b) concentration and output voltage at varying ambient temperatures in the range of 23–12 °C.
methanol solution that is exhausted out of the stack and thus maintains its temperature at a higher level. This strategy is considered effective in raising the overall energy efficiency (gsystem ) of the DMFC system because reducing the cooling rate of the heat exchanger minimizes the parasitic power loss by cooling fans. 3.2. Strategies to control the methanol concentration of DMFC systems under varying ambient temperatures 3.2.1. When the ambient temperature is lower than the reference temperature Additional experiments have been conducted to tackle the problems caused by the variations in ambient temperature. The ambient temperatures could be classified as either lower or higher than the reference temperature, which in this case is 23 °C. First, the lower range was tested by varying the cooling rate of the heat exchanger instead of varying the methanol concentration in the feed. In these experiments, the predetermined target temperature of the stack was set at 67 °C and the target range was 67 ± 1 °C. As the ambient temperature falls below the reference temperature of 23 °C after 12 min, the stack temperature drops and deviates from the target range (indicated with dotted lines) after 20 min, as shown in Fig. 3a. As the stack temperature drops, the methanol consumption rate is lowered due to a decline in the methanol crossover rate, which eventually increases the methanol concentration in the feed because the pumping rate of pure methanol to the methanol mixing chamber was kept invariant regardless of the ambient temperature. In order to maintain the stack temperature within the target range, the cooling rate of the heat exchanger was gradually lowered by decreasing the input power to the cooling fan from 2 to
107
Fig. 3. Profiles of the changes in (a) the temperatures and the concentration, and (b) output power and total system efficiency at various points when the ambient temperature is lower than 23 °C.
0.3 W at 35 min, which led to a gradual increase in the anode inlet temperature, Tan,in, and finally the stack temperature returned to within the target range of 67 ± 1 °C (45–80 min). In addition, the methanol concentration was again decreased to around the target value (3.2 wt.%) at 70–80 min. In order to compare the total system efficiencies (gsystem ) before and after the variation of the cooling rate, the output power and the gsystem were measured as shown in Fig. 3b. The gsystem is the ratio between the generated electrical energy (We, a maximum value of the electrical work that can be obtained) and the total chemical energy (Wch.e.) of the methanol consumed [20]. The relationship is given by Eq. (3).
gsystem ¼ 100
We t fðP gross gDC—DC Þ Paux g ¼ t ðV MeOH qMeOH Þ W che
ð3Þ
In Eq. (3), Pgross is the gross power output (W), gDC—DC is the DC–DC converter efficiency (%), Paux is the auxiliary power (W) consumed by the balance of plant (BOP), t is the running time (h), VMeOH is the total amount of methanol consumed (L), and qMeOH is the volumetric energy density of methanol (WL1). In this DMFC system, the efficiency of the DC–DC converter is 90%, and, thus, 5.20 Wh (1/6 31.20 W) of electric power is consumed by the auxiliary parts (BOP) to run the DMFC system. The DMFC stack generates 211.6 W during the run for 10 min (1/6 h) from 0 to 10 min, corresponding to an energy value of 31.7 Wh (1/ 6 211.6 0.9). Then, the electrical energy available (We) becomes 31.7 5.20 = 26.5 Wh. The total amount of pure methanol consumed is 0.0265 L over the total span of time, of which total chemical energy is 126.70 Wh (0.0265 L 4780 Wh L1). Eventually, the energy efficiency of the DMFC system (gsystem is 20.93% (100% 26.53/126.70) between 0 and 10 min. After the heat exchanger cooling rate is reduced to 0.3 W, the gsystem is increased
108
M.-G. An et al. / Applied Energy 129 (2014) 104–111
0.1% in the time span of 70–80 min. This increment in the gsystem may not seem significant in this case. However, it has at least confirmed that the sensor-less control of the methanol concentration could improve the efficiency of DMFC systems under varying ambient temperatures (Tam) by employing a proper control strategy. Another approach that can be used to handle the stack temperature decline due to lowered ambient temperature would be to increase the methanol concentration in the feed. The pumping rate of the pure methanol was raised stepwise by 20 mg min1 every 5 min when the stack temperature fell below the target range after 20 min, as shown in Fig. 4a. The methanol pumping rate stops increasing when the stack temperature returns to within the target range after 45 min. The value of Tan.in remains relatively low, at approximately 40 °C (30–70 min), because of the lowered ambient temperature, even though the methanol concentration is increased from 3.2 to 3.6 wt.%, as shown in Fig. 4b. In this case, the gsystem is decreased from 20.7% (0–10 min) to 18.9% (60–70 min) because of the increased pumping rate of pure methanol that leads to an increased loss of methanol. When comparing the two approaches to the restoration of the stack temperature upon decreased ambient temperature, the first approach that reduces the cooling rate of the heat exchanger is better in terms of overall energy efficiency and is easier to control than the second one that increases the methanol concentration in the feed. 3.2.2. When the ambient temperature is higher than the reference temperature When the ambient temperature, Tam, is higher than the reference temperature (23 °C), the stack temperature (Tan,out) increases because of a lower heat dissipation rate at the stack and a resultant higher methanol crossover rate. There are also two approaches to
Fig. 4. Profiles of the changes in (a) the temperatures and pure methanol pumping rates, and (b) output power, concentration, and total system efficiency when the ambient temperature is lower than 23 °C.
control the stack temperature in this situation: either increase the cooling rate of the heat exchanger or decrease the methanol concentration in the feed. When using the first approach, the cooling rate of the heat exchanger was increased to lower the temperature of the methanol solution circulating through the stack. As the ambient temperature rises from 23 to 32 °C, the stack temperature begins to increase at 17 min and crosses the upper boundaries of the target range after 35 min, as shown in Fig. 5a. Then the cooling rate of the heat exchanger is raised from 2.0 to 3.6 W, restraining further temperature rise and maintaining the stack temperature within the target range. In this case, the gsystem declines from 21.0% (0–10 min) to 20.8% (70–80 min), as shown in Fig. 6b. The second approach to control the stack temperature rise under an increase in ambient temperature involved lowering the methanol concentration in the feed. As shown in Fig. 6a, when the stack temperature increases and crosses the upper boundaries of the target range at 39 min, the pumping rate of the pure methanol is lowered stepwise by 20 mg min1 every 5 min. The methanol concentration steadily declines from 3.2 to 2.9 wt.%. Despite the decreased methanol concentration, the output power of the DMFC stack remains almost the same as that of the initial value, as shown in Fig. 6b. As a result, the gsystem is increased from 21.2% to 22.3%. When comparing the results of the two approaches to control the stack temperature under increased ambient temperatures, it is more efficient to lower the methanol concentration in the feed rather than increasing the cooling rate of the heat exchanger. It is apparent that successful control of the stack temperature can be achieved by manipulating either the heat exchanger cooling rate or the methanol concentration in the feed depending on the
Fig. 5. Profiles of the changes in (a) the temperatures and concentration, and (b) output power and total system efficiency when the ambient temperature is higher than 23 °C.
M.-G. An et al. / Applied Energy 129 (2014) 104–111
109
Based on these assessments of the effects of ambient temperature on the behaviors of a DMFC system, a new sensor-less methanol concentration control algorithm employing feedback from the ambient temperature (SLCCFA) has been devised, as explained in the following section. 3.3. Designing a SLCCFA algorithm
Fig. 6. Profiles of the changes in (a) the temperatures and methanol pumping rates, and (b) output power, concentration, and total system efficiency when the ambient temperature is higher than 23 °C.
ambient temperature: a decrease in stack temperature due to a decline in ambient temperature can be efficiently recovered by decreasing the cooling rate of the heat exchanger, and, conversely, an increase in the stack temperature due to increases in the ambient temperature can be controlled by lowering the methanol concentration in the feed.
A SLCCFA algorithm has been devised in order to control the stack temperature and methanol concentration in the feed when there is a change in the ambient temperature, as shown in Fig. 7. In the first step, the operating conditions of the DMFC stack are established in order to calculate the methanol consumption rates based on the database, Nm,t [15]. Thereafter, the stack temperature is measured in order to calculate the temperature difference with a target (set) temperature. The next step is to judge whether the ambient temperature is higher than the reference value (23 °C) and to select a proper proportional-integral (PI) controller between PI controllers 1 and 2. PI controller 1 controls the cooling rate of the anode heat exchanger, and PI controller 2 adjusts the pumping rate of pure methanol. These two PI controllers operate separately depending on the ambient temperature (Tam). If Tam is lower than 23 °C, and thus the stack temperature is lower than the set value, the stack temperature is raised by lowering the cooling rate of the anode heat exchanger, which is controlled by PI controller 1 while a constant amount of pure methanol, as determined by Nm,t, is supplied to the methanol mixing chamber. On the contrary, if the Tam is higher than 23 °C, and thus the stack temperature is higher than the set value, the stack temperature is lowered by decreasing the pumping rate of pure methanol to the methanol mixing chamber in order to decrease the methanol concentration in the feed: the pumping rate is determined by PI controller 2 based on the difference in the present stack temperature with the target value while the heat exchanger cooling rate remains invariant. The PI controllers for the heat exchanger and the puremethanol pump have different control parameters of proportional gain (KC) and integral time (sI ). There are several methods [25] available for tuning the PI control parameters, but in this study,
Fig. 7. Sensor-less methanol concentration control algorithm based on the feedback of ambient temperature (SLCCFA).
110
M.-G. An et al. / Applied Energy 129 (2014) 104–111
Fig. 8. Profiles of the changes in (a) temperatures and methanol concentration, (b) methanol and heat exchanger cooling rates, and (c) output power and total system efficiency under operation at 25 A when the ambient temperature is higher than 23 °C.
the Ziegler–Nichols method [26] was used. In addition, during the stack operation, the output voltages of the stack are measured to determine whether it is lower than a minimum set value (Vmin). If the stack voltage drops lower than the Vmin (ex., 6 V (0.3 V per each cell) in this study) and declines continuously for any reason, the DMFC system is automatically shut down to prevent any damage to the DMFC system.
3.4. Evaluation of the SLCCFA algorithm at varying ambient temperatures The SLCCFA algorithm was programmed and installed in a LabView system, and its performance to control the methanol concentration has been evaluated using a 200 W DMFC system under a constant load of 25 A and a stoichiometry of 3 for both air and methanol feed while increasing the ambient temperature from 22 to 34 °C. Target values of the stack temperature and the
Fig. 9. Profiles of the changes in (a) temperatures and the methanol concentration, (b) methanol and heat exchanger cooling rates, and (c) output power and total system efficiency under operation at 25 A when the ambient temperature is lower than 23 °C.
methanol concentration were set at 67 °C and 3.2 wt.%, respectively. Fig. 8a shows the profiles of the temperatures and methanol concentration of the stack that were controlled with the SLCCFA algorithm when the ambient temperature (Tam) was higher than 23 °C. At first, the stack temperature is maintained at approximately the set value (67 °C) in the time span of 0–50 min under a steady-state operation. Afterwards, as Tam is rapidly increased from 22 to 34 °C (see the inset), the stack temperature and the anode inlet temperature (Tan.in) rise immediately, which triggers the PI controller to lower the pumping rate of the pure methanol to reduce the methanol concentration in the feed until the stack temperature returns to the target value of 67 °C. Fig. 8b shows the changes in the pure methanol pumping rates and in the heat exchanger cooling rates when Tam starts to increase after 50 min. Under steady-state operating conditions for the first 50 min, the
M.-G. An et al. / Applied Energy 129 (2014) 104–111
stack temperature is controlled by adjusting the cooling rate of the anode heat exchanger while the pumping rate of pure methanol is kept constant. When the ambient temperature rises to 34 °C and causes the stack temperature to rise above the target value after 50 min, the stack temperature is controlled solely by adjusting the methanol pumping rate, whereas the cooling rate remains constant. Fig. 8c shows the profiles of the output power and the gsystem during the operating span mentioned above. Though the output power of the stack declines from 211.5 to 210.1 W on average over the first and the last 20 min, respectively, the energy efficiency of the DMFC system increases from 20.9% to 22.7%. The increased efficiency is due to the reduced methanol loss since the methanol concentration in the feed is lowered from 3.2 to 2.6 wt.%. When the ambient temperature drops below the reference temperature of 23 °C, the stack temperature can be controlled by adjusting the cooling rate of the anode heat exchanger instead of manipulating the methanol concentration. Fig. 9a shows the profiles of the temperatures and the methanol concentrations when the Tam declines to 15 °C. In the beginning, the stack temperature is maintained at around 67 °C over the initial 30 min. As the Tam is decreased from 22 to 15 °C (see the inset in Fig. 9a), the stack temperature also declines substantially. Then the heat exchanger begins to lower its cooling rate to restore the stack temperature while the pure methanol pumping rate remains constant, as shown in Fig. 9b. In this case, the lowered cooling rate dissipates less heat from the methanol solution that is exhausted out of the stack, and, thus, the temperature of the methanol solution is maintained at a relatively higher level. This strategy is effective in elevating the overall energy efficiency (gsystem ) of the DMFC system because the parasitic power loss by the cooling fans is minimized. Fig. 9c shows the profiles of the corresponding output power and the gsystem . The output power between 70 and 90 min is higher than that from 0 to 20 min. Therefore, the gsystem is increased from 20.9% to 21.2% because of the reduction in the parasitic power losses. 4. Conclusions In this study, a new version of an algorithm for sensor-less control of a methanol concentration was proposed and tested using a 200 W DMFC system, which could be adapted to efficiently tackle sudden changes in the ambient temperature. When the ambient temperature is higher than the reference temperature, the stack temperature is controlled by decreasing the methanol concentration in the feed, which is implemented by decreasing the pumping rate of pure methanol to the methanol mixing chamber. By using this method, the total energy efficiency of the DMFC system is improved from 20.9% to 22.7%. A lower methanol concentration results in lower methanol crossover and lower heat generation at the cathode, leading to a lower methanol loss and thus higher energy efficiency. Conversely, when the ambient temperature is lower than 23 °C, the lowered stack temperature is controlled by reducing the heat exchanger cooling rate of the methanol solution circulating through the stack to compensate for the heat loss of the DMFC stack. Through this method, the stack temperature is maintained at a target value, and the energy efficiency of the DMFC system is increased from 20.9% to 21.2%. This new algorithm enables efficient and indirect control of the methanol concentration of DMFC systems through control of the stack temperature when there is a noticeable change in the ambient temperature. The sensor-less methanol concentration controller has no lifetime limitation issues. Thus, this new version would provide an adequate and inexpensive solution for controlling the methanol concentration and would help to reduce the cost of commercial DMFC systems.
111
Acknowledgments This work was supported by the Korea Ministry of Trade, Industry, and Energy through the Korea Institute of Energy Technology Evaluation and Planning (KETEP) under Grant No. 2011-30300-30010, and was also supported by the Korea CCS R&D Center (KCRC) Grant funded by the Korea government (Ministry of Science, ICT and Future) (No. 2012-0008931). References [1] Kim J-H, Yang M-J, Park J-Y. Improvement on performance and efficiency of direct methanol fuel cells using hydrocarbon-based membrane electrode assembly. Appl Energy 2014;115:95–102. [2] Wang S, Sun G, Wang G, Zhou Z, Zhao X, Sun H, et al. Improvement of direct methanol fuel cell performance by modifying catalyst coated membrane structure. Electrochem Commun 2005;7:1007–12. [3] Achmad F, Kamarudin SK, Daud WRW, Majlan EH. Passive direct methanol fuel cells for portable electronic devices. Appl Energy 2011;88:1681–9. [4] Han S, Yun Y, Park K-W, Sung Y-E, Hyeon T. Simple solid-phase synthesis of hollow graphitic nanoparticles and their application to direct methanol fuel cell electrodes. Adv Mater 2003;15:1922–5. [5] Pasupathi S, Ji S, Bladergroen BJ, Linkov V. High DMFC performance output using modified acid-based polymer blend. Int J Hydrogen Energy 2008;33:3132–6. [6] Zhang J, Yin G-P, Lai Q-Z, Wang Z-B, Cai K-D, Liu P. The influence of anode gas diffusion layer on the performance of low-temperature DMFC. J Power Sources 2007;168:453–8. [7] Zainoodin AM, Kamarudin SK, Masdar MS, Daud WRW, Mohamad AB, Sahari J. High power direct methanol fuel cell with a porous carbon nanofiber anode layer. Appl Energy 2014;113:946–54. [8] Jung H-Y, Park J-K. Long-term performance of DMFC based on the blend membrane of sulfornated poly(ether ether ketone) and poly(vinylidene fluoride). Int J Hydrogen Energy 2009;34:3915–21. [9] Mehmood A, Ha HY. Performance restoration of direct methanol fuel cells in long-term operation using a hydrogen evolution method. Appl Energy 2014;114:164–71. [10] Dohle H, Divisek J, Mergel J, Oetjen HF, Zingler C, Stolten D. Recent developments of the measurement of the methanol permeation in a direct methanol fuel cell. J Power Sources 2002;105:274–82. [11] Arico AS, Srinivasan S, Antonucci V. DMFCs: From fundamental aspects to technology development. Fuel Cells 2001;1:133–61. [12] Yang T. Composite membrane of sulfonated poly(ether ether ketone) and sulfated poly(vinyl alcohol) for use in direct methanol fuel cells. J Membr Sci 2009;35:2160–75. [13] Chiu Y-J, Lien H-C. A strategy of estimating fuel concentration in a direct liquid-feed fuel cell system. J Power Sources 2006;159:1162–8. [14] Zhao H, Shen J, Zhang J, Wang H, Wilkinson DP, Gu CE. Liquid methanol concentration sensors for direct methanol fuel cells. J Power Sources 2006;159:626–36. [15] Ha TJ, Kim J-H, Joh H-I, Kim S-K, Moon G-Y, Lim T-H, et al. Sensor-less control of methanol concentration based on estimation of methanol consumption rates for direct methanol fuel cells systems. Int J Hydrogen Energy 2008;33:7163–71. [16] Shen K-S, Wan C-C, Wang Y-Y, Yu TL, Chiu Y-J. An algorithm for sensor-less fuel control of direct methanol fuel cells. J Power Sources 2010;195:4785–95. [17] Chang CL, Chen CY, Sung CC, Liou DH. Fuel sensor-less control of a liquid feed fuel cell system under steady load for portable applications. J Power Sources 2007;164:606–13. [18] Chang CL, Chen CY, Sung CC, Liou DH. Fuel sensor-less control of a liquid feed fuel cell under dynamic loading conditions for portable power sources (I). J Power Sources 2008;182:133–40. [19] Chang CL, Chen CY, Sung CC, Liou DH, Chang CY, Cha HC. Fuel sensor-less control of a liquid feed fuel cell under dynamic loading conditions for portable power sources (II). J Power Sources 2010;195:1427–34. [20] Chen CY, Liu DH, Huang CL, Chang CL. Portable DMFC system with methanol sensor-less control. J Power Sources 2007;167:442–9. [21] Chen CY, Chang CL, Sung CC. Operation characteristic analysis of a direct methanol fuel cell system using the methanol sensor-less control method. Fuel Cells 2012;12:883–91. [22] Arisetty S, Jacob CA, Prasad AK, Advani SG. Regulating methanol feed concentration in direct methanol fuel cells using feedback from voltage measurements. J Power Sources 2009;187:415–21. [23] Lian K-Y, Yang C-M. Sensor-less adaptive fuel concentration control for direct methanol fuel cells under varying load. J Power Sources 2013;231:239–45. [24] An M-G, Mehmood A, Ha HY. (2014) A sensor-less methanol concentration control system based on the feedback of stack temperature, Appl Energy. [25] Skogestad S. Simple analytic rules for model reduction and PID controller tuning. J Process Control 2003;13:291–309. [26] Ziegler JG, Nichols NB. Optimum settings for automatic controllers. Trans ASME 1942;64:759–68.
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Sensor-less control of the methanol concentration of direct methanol fuel cells at varying ambient temperatures Myung-Gi An, Asad Mehmood, Heung Yong Ha ⇑ Energy Convergence Research Center, Korea Institute of Science and Technology (KIST), Seoul 136-791, Republic of Korea Department of Energy and Environmental Engineering, Korea University of Science and Technology (UST), Yuseong-gu, Daejeon 305-333, Republic of Korea
h i g h l i g h t s A new algorithm is proposed for the sensor-less control of methanol concentration. Two different strategies are used depending on the ambient temperatures. Energy efficiency of the DMFC system has been improved by using the new algorithm.
a r t i c l e
i n f o
Article history: Received 2 February 2014 Received in revised form 24 April 2014 Accepted 28 April 2014
Keywords: Direct methanol fuel cell Sensor-less control Methanol concentration Stack temperature Ambient temperature
a b s t r a c t A new version of an algorithm is used to control the methanol concentration in the feed of DMFC systems without using methanol sensors under varying ambient temperatures. The methanol concentration is controlled indirectly by controlling the temperature of the DMFC stack, which correlates well with the methanol concentration. Depending on the ambient temperature relative to a preset reference temperature, two different strategies are used to control the stack temperature: either reducing the cooling rate of the methanol solution passing through an anode-side heat exchanger; or, lowering the pumping rate of the pure methanol to the depleted feed solution. The feasibility of the algorithm is evaluated using a DMFC system that consists of a 200 W stack and the balance of plant (BOP). The DMFC system includes a sensor-less methanol controller that is operated using a LabView system as the central processing unit. The algorithm is experimentally confirmed to precisely control the methanol concentration and the stack temperature at target values under an environment of varying ambient temperatures. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Direct methanol fuel cells (DMFCs) continue to garner interest as a clean energy technology for powering small and portable electronic devices because of positive characteristics such as high-energy density, flexibility in power output from sub-watt to several hundred watts, and reliable operation. Over the past decade, a number of worldwide efforts have been focused on improving the performance and durability of DMFCs [1–9]. However, DMFCs still suffer from low efficiency because of poor electrode kinetics and methanol crossover from the anode to the cathode through the polymer electrolyte membrane [10,11]. The methanol crossover leads to not only fuel waste, but also mixed potential at the cathode, which adversely affects the performance and fuel ⇑ Corresponding author at: Energy Convergence Research Center, Korea Institute of Science and Technology (KIST), Seoul 136-791, Republic of Korea. Tel.: +82 2 958 5275; fax: +82 2 958 5229. E-mail address: [email protected] (H.Y. Ha). http://dx.doi.org/10.1016/j.apenergy.2014.04.100 0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.
efficiency [12]. The methanol crossover rate is mainly influenced by the concentration of the methanol feed at the anode and by the operating temperature of the DMFC [13]. Controlling the methanol concentration in an adequate range therefore reduces the methanol crossover under given operating conditions, and plays a critical role in stable and efficient DMFC operations. In order to achieve this goal, electronic sensors are generally installed in a feed re-circulating DMFC system to monitor and control the methanol concentration. However, the sensors are known to have many problems in terms of cost, size, durability and reliability. Zhao et al. [14] reviewed various methanol sensors for DMFCs that are generally classified into two groups: electrochemical and physical. Physical sensors are reliable and have a wide measurement range, but they are expensive and too bulky for use in portable DMFC systems. Electrochemical sensors are known to have a narrow sensing range and a slow response time due to diffusion limitations, which hampers their reliability. These also have durability issues due to a degradation of the electrolyte membranes and a deterioration of the electrodes with operating time. Some efforts have been made
M.-G. An et al. / Applied Energy 129 (2014) 104–111
to solve the problems of electronic sensors, and, as a result, sensorless methanol controllers have been proposed and reported by some research groups. Several researchers have reported sensor-less methanol control using consumption equations rather than sensors to estimate the methanol concentration in the feed [13,15,16]. Chiu and Lien [13] proposed a three-dimensional measurement space (CCS) and an interpolation algorithm (ICCS) based on the constant-concentration surfaces. Once the CCS (I, V, T) is obtained for the known methanol concentrations in the feed, the unknown methanol concentration that corresponds to the in situ-measured data (Iu, Vu, Tu) can be estimated by interpolation. By using the ICCS, they estimated the methanol concentrations in the feed based on the current, voltage, and cell temperature of a fuel cell. Ha et al. [15] devised a sensor-less algorithm that could control the methanol concentration at a set value by supplying the same amount of methanol consumed in a DMFC system to the recirculating methanol feed solution. They built a database of methanol consumption rates that were collected under various operating conditions and utilized the database to calculate the amount of methanol needed to maintain the methanol feed concentration at a set value. Shen et al. [16] reported a real-time fuel control algorithm based on Chiu and Lien’s ICCS (I, V, T) algorithm. They modified the ICCS algorithm by accounting for the ‘‘MEA decay’’ and including an in situ estimating method in their control program that could estimate the methanol and water consumption quantities by accumulating the operating time and the methanol and water consumption rates. The program could be used to determine the remaining amount of methanol in the operating DMFC system that could be used to control the methanol concentration by compensating for the depleted methanol feed solution. On the other hand, other studies [17–23] have reported sensor-less methods that utilized the DMFC operating characteristics as feedback parameters to control the methanol concentration in the feed. Chang et al. [17,18] reported a sensor-less algorithm that was referred to as ‘‘impulse response based on discrete-time fuel-injection’’ (IR-DTFI) along with a modified version that could regulate the fuel concentration in order to optimize fuel cell performance by accounting for the changes in the characteristic values of a DMFC stack, such as voltage, current, and power, during a set period of operation. They [19] also suggested an advanced version referred to as ‘‘impulse response based on current-integral and discrete-time fuel-injection’’ (IR-CIDTFI) algorithm that could shorten the monitoring period (5 s or less) by calculating the amount of fuel consumed during the last monitoring cycle for faster system response and greater stability. They tested the performance of IR-DTFI with a 40 W DMFC system to power portable electronics and to evaluate the effectiveness of both IRDTFI and IR-CIDTFI algorithms from the point of operating characteristics [20,21]. Arisetty et al. [22] found that maximum voltage could be obtained by adjusting the methanol concentration in the feed under a given current density. Based on this finding, they developed an in situ sensor-less methodology that employed the cell voltage as the feedback to optimize the methanol concentration for maximum power density under dynamic operating conditions while maintaining a high level of fuel utilization. Lian and Yang [23] developed a sensor-less adaptive fuel concentration control (SAFCC) algorithm to regulate the methanol concentration in a suitable range by detecting transient voltage behavior under pulse-like changes in the load current. Although those sensor-less controllers are believed to be useful in controlling the methanol concentration, no controller has ever accounted for the effect of the ambient temperature surrounding a DMFC system. Our first version of the sensor-less methanol concentration controller [15] was designed to regulate the methanol concentration in the feed by supplying an amount of pure methanol to the recirculating methanol feed solution that would be equal to the amount of
105
methanol consumed in a DMFC system. Later, the algorithm was revised to improve its accuracy by using the stack temperature as a feedback parameter, which has a close correlation with the methanol concentration [24]. This algorithm has been further modified during this study by including the ambient temperature as a feedback parameter. Ambient temperature affects the stack temperature to allow deviation from a set value. Change in the stack temperature can be recovered by adjusting either the cooling rate of the anode heat exchanger or the pumping rate of the pure methanol to the depleted feed solution. When the ambient temperature is lower than the reference temperature (i.e., 23 °C in this study), the stack temperature decreases to below the set value because of excessive heat loss from the stack. In this case, the temperature of the circulating methanol solution can be raised by decreasing the cooling rate of the heat exchanger to compensate for the heat loss from the DMFC stack while the pumping rate of the pure methanol remains unchanged. On the other hand, when the ambient temperature is higher than the reference value, however, the heat loss from the stack is lowered and therefore the stack temperature increases above the set value. In this case, the pure methanol pumping rate can be lowered to decrease the methanol concentration in the feed, thus returning the stack temperature to the set value. The lowered methanol concentration reduces the methanol crossover rate, decreasing the heat generation at the cathode and lowering the stack temperature. In the present study, a new version of a sensor-less methanol concentration control algorithm has been proposed based on the feedback from ambient temperatures (SLCCFA). The effects of ambient temperature on the temperature and the feed concentration of a DMFC stack are explored in detail under various operating conditions, and these are used in designing a sensor-less algorithm. The feasibility of the new sensor-less algorithm has been evaluated in terms of the methanol concentration, stack temperature and system efficiency of a 200 WDMFC system.
2. Experimental setup 2.1. Building a database of methanol consumption rates in a DMFC stack In this control algorithm, the methanol consumption rates measured with a typical DMFC were used to calculate the amounts of pure methanol needed to compensate for the depleted feed solution under various operating conditions. The amounts of pure methanol to be pumped were determined by the experimental equations made in our earlier work [15]. The methanol consumption rates had been measured using a large-size single-cell DMFC with an active area of 150 cm2 under various operating conditions by changing the output current, methanol concentration in the feed (1.3, 2.6, 3.2, and 3.8 wt.%), and cell temperature (40, 60, and 80 °C) under fixed flow rates of air (1118 ml min1) and methanol feed (8.78 ml min1) based on a 3/3 stoichiometry (O2/ methanol). The total methanol consumption rate (Nm,t) in a DMFC is the sum of the methanol consumption rates by electrochemical oxidation (Nm,e) and methanol crossover (Nm,x) (Eq. (1)) [15].
Nm;t ¼ Nm;e þ Nm;x
ð1Þ
Nm,e was estimated based on the output current, and Nm,x was obtained by measuring the amount of CO2 at the cathode outlet that was produced by the oxidation of crossed-over methanol at the cathode. The amount of unreacted methanol that was exhausted from the cathode outlet was considered negligible because most of the crossed-over methanol was oxidized to CO2 at the cathode. However, a significant amount of CO2 generated
106
M.-G. An et al. / Applied Energy 129 (2014) 104–111
by the electrochemical oxidation of methanol at the anode penetrated through the membrane into the cathode. Therefore, Nm,x was estimated by the total CO2 flux measured at the cathode outlet ðN mCO2 Þ minus the CO2 crossover ðN xCO2 Þ (Eq. (2)). To measure the crossed-over CO2 from the anode to the cathode, a half-cell test was carried out using a fixed flow rate of nitrogen that equaled that of the air used for full-cell tests.
Nm;x ¼ NmCO2 NxCO2
catalyst on the anode and a 2 mg-Pt cm2 catalyst on the cathode. Graphite plates with a parallel serpentine-type flow field were used as separators. The flow rates of the methanol solution supplied to the anode, and the dry air supplied to the cathode, were set at 240 ml min1 and 23 L min1, respectively. The DMFC stack was operated on a constant electric load of 25 A (166 mA cm2) at a temperature of approximately 67 °C.
ð2Þ
The Nm,t was then calculated for the DMFC. In order to calculate the Nm,t for the 20-cell DMFC stack, it was multiplied by 20, and used for the operation of a DMFC stack under a constant electric load of 25 A. 2.2. Configuration of a sensor-less DMFC system The DMFC system was comprised of a methanol mixing chamber, three liquid pumps, a 200 W stack, two heat exchangers with air cooling fans, a pure methanol reservoir, a water reservoir, a liquid-level sensor, several thermocouples (K-type), and an air blower to the cathode, as shown in Fig. 1. The DMFC system was also equipped with a SLCCFA processor and a DC–DC converter. The SLCCFA processor simultaneously functioned as a central processing unit that regulated the operations of all the components in the DMFC system. Three liquid pumps were used: The first (a puremethanol pump) for pumping pure methanol into the methanol mixing chamber, the second (a main pump) for circulating a methanol feed solution along the recirculation loop passing through the DMFC stack, and the third (a water pump) for supplying condensed water from the cathode heat exchanger to the methanol mixing chamber. When the level of the methanol solution in the methanol-mixing chamber fell below the preset height, the water-feed pump started supplying water from the water reservoir to the mixing chamber. There were three thermo-couples (K-type) attached to the inlet (Tan.in) and outlet (T) of the anode side of the stack and the outlet (Th.out) of the anode heat exchanger to measure the temperatures of the methanol feed circulating along the fuel loop. The anode outlet temperature (T) was considered to be the temperature of the stack. In order to measure the methanol concentration, a methanol sensor (FC 10, ISSYS) was installed between the mixing chamber and the stack, which was used only for monitoring but not for controlling the methanol concentration. Each air-cooled heat exchanger made of stainless steel was integrated with a cooling fan. 2.3. DMFC stack specifications and test conditions The 200 W-class DMFC stack consisted of 20 single-cells with MEAs that had an active area of 150 cm2. The MEAs were purchased from Johnson-Matthey, Inc. and were made of Nafion 115 membranes and carbon paper electrodes with a 6 mg-PtRu cm2
3. Results and discussion 3.1. Behavior of the 200 W DMFC system at varying ambient temperatures First, background data were collected in order to design a sensor-less methanol concentration control algorithm. The experiments were carried out to evaluate the effects of ambient temperature on the behavior of the DMFC system. In these experiments, the DMFC system was operated at a steady state and the methanol concentration in the feed was controlled by supplying a fixed amount of pure methanol (2.1 g min1) to the recirculating methanol feed solution that was estimated to be consumed under given operating conditions of an electric load of 25 A, a stack temperature of 67 °C, a methanol concentration of 1 M (3.2 wt.%), and a stoichiometry (k) of both reactants at 3 (the flow rates of methanol solution and dry air were set at 240 ml min1 and 23 L min1, respectively) [15]. Fig. 2 shows the changes in the temperature, methanol feed concentration and output voltage of the DMFC stack as the ambient temperature (Tam) varies in the range of 12.5– 23.0 °C. The temperatures of the methanol solution were measured at the anode inlet (Tan.in) and anode outlet (T, considered equal to the stack temperature) of the stack and at the heat exchanger outlet (Th.out). In the beginning stage of the run from 0 to 40 min in Fig. 2a, the Tam is 23 °C and the stack temperature is maintained almost constant. When the Tam starts to decrease from 23.0 to 12.5 °C (at 40 min), the stack temperature decreases remarkably followed by drops in Th.out and Tan.in. The lowered stack temperature reduces the methanol consumption rate due to the reduced methanol crossover from the anode to the cathode and thus results in an increased methanol concentration in the feed from 3.2 to 3.4 wt.%, as shown in Fig. 2b. Afterwards, as the Tam increases again after 100 min, the stack temperature and the methanol concentration gradually return back to their initial values. In Fig. 2b, the output voltage also decreases corresponding to the decreased Tam even though the methanol concentration is increased above the target value in a time span of 60–120 min. This result shows that simply increasing the methanol feed concentration may not compensate for the decreased performance due to a lowered stack temperature. In this case, it may be better to seek a method that will deter a drop in the stack temperature. This could be achieved by lowering the cooling rate of the heat exchanger on the methanol circulation loop: the lowered cooling rate dissipates less heat from the
Fig. 1. Configuration of a sensor-less DMFC system with methanol re-circulation.
M.-G. An et al. / Applied Energy 129 (2014) 104–111
Fig. 2. Profiles of the changes in (a) the temperatures, and (b) concentration and output voltage at varying ambient temperatures in the range of 23–12 °C.
methanol solution that is exhausted out of the stack and thus maintains its temperature at a higher level. This strategy is considered effective in raising the overall energy efficiency (gsystem ) of the DMFC system because reducing the cooling rate of the heat exchanger minimizes the parasitic power loss by cooling fans. 3.2. Strategies to control the methanol concentration of DMFC systems under varying ambient temperatures 3.2.1. When the ambient temperature is lower than the reference temperature Additional experiments have been conducted to tackle the problems caused by the variations in ambient temperature. The ambient temperatures could be classified as either lower or higher than the reference temperature, which in this case is 23 °C. First, the lower range was tested by varying the cooling rate of the heat exchanger instead of varying the methanol concentration in the feed. In these experiments, the predetermined target temperature of the stack was set at 67 °C and the target range was 67 ± 1 °C. As the ambient temperature falls below the reference temperature of 23 °C after 12 min, the stack temperature drops and deviates from the target range (indicated with dotted lines) after 20 min, as shown in Fig. 3a. As the stack temperature drops, the methanol consumption rate is lowered due to a decline in the methanol crossover rate, which eventually increases the methanol concentration in the feed because the pumping rate of pure methanol to the methanol mixing chamber was kept invariant regardless of the ambient temperature. In order to maintain the stack temperature within the target range, the cooling rate of the heat exchanger was gradually lowered by decreasing the input power to the cooling fan from 2 to
107
Fig. 3. Profiles of the changes in (a) the temperatures and the concentration, and (b) output power and total system efficiency at various points when the ambient temperature is lower than 23 °C.
0.3 W at 35 min, which led to a gradual increase in the anode inlet temperature, Tan,in, and finally the stack temperature returned to within the target range of 67 ± 1 °C (45–80 min). In addition, the methanol concentration was again decreased to around the target value (3.2 wt.%) at 70–80 min. In order to compare the total system efficiencies (gsystem ) before and after the variation of the cooling rate, the output power and the gsystem were measured as shown in Fig. 3b. The gsystem is the ratio between the generated electrical energy (We, a maximum value of the electrical work that can be obtained) and the total chemical energy (Wch.e.) of the methanol consumed [20]. The relationship is given by Eq. (3).
gsystem ¼ 100
We t fðP gross gDC—DC Þ Paux g ¼ t ðV MeOH qMeOH Þ W che
ð3Þ
In Eq. (3), Pgross is the gross power output (W), gDC—DC is the DC–DC converter efficiency (%), Paux is the auxiliary power (W) consumed by the balance of plant (BOP), t is the running time (h), VMeOH is the total amount of methanol consumed (L), and qMeOH is the volumetric energy density of methanol (WL1). In this DMFC system, the efficiency of the DC–DC converter is 90%, and, thus, 5.20 Wh (1/6 31.20 W) of electric power is consumed by the auxiliary parts (BOP) to run the DMFC system. The DMFC stack generates 211.6 W during the run for 10 min (1/6 h) from 0 to 10 min, corresponding to an energy value of 31.7 Wh (1/ 6 211.6 0.9). Then, the electrical energy available (We) becomes 31.7 5.20 = 26.5 Wh. The total amount of pure methanol consumed is 0.0265 L over the total span of time, of which total chemical energy is 126.70 Wh (0.0265 L 4780 Wh L1). Eventually, the energy efficiency of the DMFC system (gsystem is 20.93% (100% 26.53/126.70) between 0 and 10 min. After the heat exchanger cooling rate is reduced to 0.3 W, the gsystem is increased
108
M.-G. An et al. / Applied Energy 129 (2014) 104–111
0.1% in the time span of 70–80 min. This increment in the gsystem may not seem significant in this case. However, it has at least confirmed that the sensor-less control of the methanol concentration could improve the efficiency of DMFC systems under varying ambient temperatures (Tam) by employing a proper control strategy. Another approach that can be used to handle the stack temperature decline due to lowered ambient temperature would be to increase the methanol concentration in the feed. The pumping rate of the pure methanol was raised stepwise by 20 mg min1 every 5 min when the stack temperature fell below the target range after 20 min, as shown in Fig. 4a. The methanol pumping rate stops increasing when the stack temperature returns to within the target range after 45 min. The value of Tan.in remains relatively low, at approximately 40 °C (30–70 min), because of the lowered ambient temperature, even though the methanol concentration is increased from 3.2 to 3.6 wt.%, as shown in Fig. 4b. In this case, the gsystem is decreased from 20.7% (0–10 min) to 18.9% (60–70 min) because of the increased pumping rate of pure methanol that leads to an increased loss of methanol. When comparing the two approaches to the restoration of the stack temperature upon decreased ambient temperature, the first approach that reduces the cooling rate of the heat exchanger is better in terms of overall energy efficiency and is easier to control than the second one that increases the methanol concentration in the feed. 3.2.2. When the ambient temperature is higher than the reference temperature When the ambient temperature, Tam, is higher than the reference temperature (23 °C), the stack temperature (Tan,out) increases because of a lower heat dissipation rate at the stack and a resultant higher methanol crossover rate. There are also two approaches to
Fig. 4. Profiles of the changes in (a) the temperatures and pure methanol pumping rates, and (b) output power, concentration, and total system efficiency when the ambient temperature is lower than 23 °C.
control the stack temperature in this situation: either increase the cooling rate of the heat exchanger or decrease the methanol concentration in the feed. When using the first approach, the cooling rate of the heat exchanger was increased to lower the temperature of the methanol solution circulating through the stack. As the ambient temperature rises from 23 to 32 °C, the stack temperature begins to increase at 17 min and crosses the upper boundaries of the target range after 35 min, as shown in Fig. 5a. Then the cooling rate of the heat exchanger is raised from 2.0 to 3.6 W, restraining further temperature rise and maintaining the stack temperature within the target range. In this case, the gsystem declines from 21.0% (0–10 min) to 20.8% (70–80 min), as shown in Fig. 6b. The second approach to control the stack temperature rise under an increase in ambient temperature involved lowering the methanol concentration in the feed. As shown in Fig. 6a, when the stack temperature increases and crosses the upper boundaries of the target range at 39 min, the pumping rate of the pure methanol is lowered stepwise by 20 mg min1 every 5 min. The methanol concentration steadily declines from 3.2 to 2.9 wt.%. Despite the decreased methanol concentration, the output power of the DMFC stack remains almost the same as that of the initial value, as shown in Fig. 6b. As a result, the gsystem is increased from 21.2% to 22.3%. When comparing the results of the two approaches to control the stack temperature under increased ambient temperatures, it is more efficient to lower the methanol concentration in the feed rather than increasing the cooling rate of the heat exchanger. It is apparent that successful control of the stack temperature can be achieved by manipulating either the heat exchanger cooling rate or the methanol concentration in the feed depending on the
Fig. 5. Profiles of the changes in (a) the temperatures and concentration, and (b) output power and total system efficiency when the ambient temperature is higher than 23 °C.
M.-G. An et al. / Applied Energy 129 (2014) 104–111
109
Based on these assessments of the effects of ambient temperature on the behaviors of a DMFC system, a new sensor-less methanol concentration control algorithm employing feedback from the ambient temperature (SLCCFA) has been devised, as explained in the following section. 3.3. Designing a SLCCFA algorithm
Fig. 6. Profiles of the changes in (a) the temperatures and methanol pumping rates, and (b) output power, concentration, and total system efficiency when the ambient temperature is higher than 23 °C.
ambient temperature: a decrease in stack temperature due to a decline in ambient temperature can be efficiently recovered by decreasing the cooling rate of the heat exchanger, and, conversely, an increase in the stack temperature due to increases in the ambient temperature can be controlled by lowering the methanol concentration in the feed.
A SLCCFA algorithm has been devised in order to control the stack temperature and methanol concentration in the feed when there is a change in the ambient temperature, as shown in Fig. 7. In the first step, the operating conditions of the DMFC stack are established in order to calculate the methanol consumption rates based on the database, Nm,t [15]. Thereafter, the stack temperature is measured in order to calculate the temperature difference with a target (set) temperature. The next step is to judge whether the ambient temperature is higher than the reference value (23 °C) and to select a proper proportional-integral (PI) controller between PI controllers 1 and 2. PI controller 1 controls the cooling rate of the anode heat exchanger, and PI controller 2 adjusts the pumping rate of pure methanol. These two PI controllers operate separately depending on the ambient temperature (Tam). If Tam is lower than 23 °C, and thus the stack temperature is lower than the set value, the stack temperature is raised by lowering the cooling rate of the anode heat exchanger, which is controlled by PI controller 1 while a constant amount of pure methanol, as determined by Nm,t, is supplied to the methanol mixing chamber. On the contrary, if the Tam is higher than 23 °C, and thus the stack temperature is higher than the set value, the stack temperature is lowered by decreasing the pumping rate of pure methanol to the methanol mixing chamber in order to decrease the methanol concentration in the feed: the pumping rate is determined by PI controller 2 based on the difference in the present stack temperature with the target value while the heat exchanger cooling rate remains invariant. The PI controllers for the heat exchanger and the puremethanol pump have different control parameters of proportional gain (KC) and integral time (sI ). There are several methods [25] available for tuning the PI control parameters, but in this study,
Fig. 7. Sensor-less methanol concentration control algorithm based on the feedback of ambient temperature (SLCCFA).
110
M.-G. An et al. / Applied Energy 129 (2014) 104–111
Fig. 8. Profiles of the changes in (a) temperatures and methanol concentration, (b) methanol and heat exchanger cooling rates, and (c) output power and total system efficiency under operation at 25 A when the ambient temperature is higher than 23 °C.
the Ziegler–Nichols method [26] was used. In addition, during the stack operation, the output voltages of the stack are measured to determine whether it is lower than a minimum set value (Vmin). If the stack voltage drops lower than the Vmin (ex., 6 V (0.3 V per each cell) in this study) and declines continuously for any reason, the DMFC system is automatically shut down to prevent any damage to the DMFC system.
3.4. Evaluation of the SLCCFA algorithm at varying ambient temperatures The SLCCFA algorithm was programmed and installed in a LabView system, and its performance to control the methanol concentration has been evaluated using a 200 W DMFC system under a constant load of 25 A and a stoichiometry of 3 for both air and methanol feed while increasing the ambient temperature from 22 to 34 °C. Target values of the stack temperature and the
Fig. 9. Profiles of the changes in (a) temperatures and the methanol concentration, (b) methanol and heat exchanger cooling rates, and (c) output power and total system efficiency under operation at 25 A when the ambient temperature is lower than 23 °C.
methanol concentration were set at 67 °C and 3.2 wt.%, respectively. Fig. 8a shows the profiles of the temperatures and methanol concentration of the stack that were controlled with the SLCCFA algorithm when the ambient temperature (Tam) was higher than 23 °C. At first, the stack temperature is maintained at approximately the set value (67 °C) in the time span of 0–50 min under a steady-state operation. Afterwards, as Tam is rapidly increased from 22 to 34 °C (see the inset), the stack temperature and the anode inlet temperature (Tan.in) rise immediately, which triggers the PI controller to lower the pumping rate of the pure methanol to reduce the methanol concentration in the feed until the stack temperature returns to the target value of 67 °C. Fig. 8b shows the changes in the pure methanol pumping rates and in the heat exchanger cooling rates when Tam starts to increase after 50 min. Under steady-state operating conditions for the first 50 min, the
M.-G. An et al. / Applied Energy 129 (2014) 104–111
stack temperature is controlled by adjusting the cooling rate of the anode heat exchanger while the pumping rate of pure methanol is kept constant. When the ambient temperature rises to 34 °C and causes the stack temperature to rise above the target value after 50 min, the stack temperature is controlled solely by adjusting the methanol pumping rate, whereas the cooling rate remains constant. Fig. 8c shows the profiles of the output power and the gsystem during the operating span mentioned above. Though the output power of the stack declines from 211.5 to 210.1 W on average over the first and the last 20 min, respectively, the energy efficiency of the DMFC system increases from 20.9% to 22.7%. The increased efficiency is due to the reduced methanol loss since the methanol concentration in the feed is lowered from 3.2 to 2.6 wt.%. When the ambient temperature drops below the reference temperature of 23 °C, the stack temperature can be controlled by adjusting the cooling rate of the anode heat exchanger instead of manipulating the methanol concentration. Fig. 9a shows the profiles of the temperatures and the methanol concentrations when the Tam declines to 15 °C. In the beginning, the stack temperature is maintained at around 67 °C over the initial 30 min. As the Tam is decreased from 22 to 15 °C (see the inset in Fig. 9a), the stack temperature also declines substantially. Then the heat exchanger begins to lower its cooling rate to restore the stack temperature while the pure methanol pumping rate remains constant, as shown in Fig. 9b. In this case, the lowered cooling rate dissipates less heat from the methanol solution that is exhausted out of the stack, and, thus, the temperature of the methanol solution is maintained at a relatively higher level. This strategy is effective in elevating the overall energy efficiency (gsystem ) of the DMFC system because the parasitic power loss by the cooling fans is minimized. Fig. 9c shows the profiles of the corresponding output power and the gsystem . The output power between 70 and 90 min is higher than that from 0 to 20 min. Therefore, the gsystem is increased from 20.9% to 21.2% because of the reduction in the parasitic power losses. 4. Conclusions In this study, a new version of an algorithm for sensor-less control of a methanol concentration was proposed and tested using a 200 W DMFC system, which could be adapted to efficiently tackle sudden changes in the ambient temperature. When the ambient temperature is higher than the reference temperature, the stack temperature is controlled by decreasing the methanol concentration in the feed, which is implemented by decreasing the pumping rate of pure methanol to the methanol mixing chamber. By using this method, the total energy efficiency of the DMFC system is improved from 20.9% to 22.7%. A lower methanol concentration results in lower methanol crossover and lower heat generation at the cathode, leading to a lower methanol loss and thus higher energy efficiency. Conversely, when the ambient temperature is lower than 23 °C, the lowered stack temperature is controlled by reducing the heat exchanger cooling rate of the methanol solution circulating through the stack to compensate for the heat loss of the DMFC stack. Through this method, the stack temperature is maintained at a target value, and the energy efficiency of the DMFC system is increased from 20.9% to 21.2%. This new algorithm enables efficient and indirect control of the methanol concentration of DMFC systems through control of the stack temperature when there is a noticeable change in the ambient temperature. The sensor-less methanol concentration controller has no lifetime limitation issues. Thus, this new version would provide an adequate and inexpensive solution for controlling the methanol concentration and would help to reduce the cost of commercial DMFC systems.
111
Acknowledgments This work was supported by the Korea Ministry of Trade, Industry, and Energy through the Korea Institute of Energy Technology Evaluation and Planning (KETEP) under Grant No. 2011-30300-30010, and was also supported by the Korea CCS R&D Center (KCRC) Grant funded by the Korea government (Ministry of Science, ICT and Future) (No. 2012-0008931). References [1] Kim J-H, Yang M-J, Park J-Y. Improvement on performance and efficiency of direct methanol fuel cells using hydrocarbon-based membrane electrode assembly. Appl Energy 2014;115:95–102. [2] Wang S, Sun G, Wang G, Zhou Z, Zhao X, Sun H, et al. Improvement of direct methanol fuel cell performance by modifying catalyst coated membrane structure. Electrochem Commun 2005;7:1007–12. [3] Achmad F, Kamarudin SK, Daud WRW, Majlan EH. Passive direct methanol fuel cells for portable electronic devices. Appl Energy 2011;88:1681–9. [4] Han S, Yun Y, Park K-W, Sung Y-E, Hyeon T. Simple solid-phase synthesis of hollow graphitic nanoparticles and their application to direct methanol fuel cell electrodes. Adv Mater 2003;15:1922–5. [5] Pasupathi S, Ji S, Bladergroen BJ, Linkov V. High DMFC performance output using modified acid-based polymer blend. Int J Hydrogen Energy 2008;33:3132–6. [6] Zhang J, Yin G-P, Lai Q-Z, Wang Z-B, Cai K-D, Liu P. The influence of anode gas diffusion layer on the performance of low-temperature DMFC. J Power Sources 2007;168:453–8. [7] Zainoodin AM, Kamarudin SK, Masdar MS, Daud WRW, Mohamad AB, Sahari J. High power direct methanol fuel cell with a porous carbon nanofiber anode layer. Appl Energy 2014;113:946–54. [8] Jung H-Y, Park J-K. Long-term performance of DMFC based on the blend membrane of sulfornated poly(ether ether ketone) and poly(vinylidene fluoride). Int J Hydrogen Energy 2009;34:3915–21. [9] Mehmood A, Ha HY. Performance restoration of direct methanol fuel cells in long-term operation using a hydrogen evolution method. Appl Energy 2014;114:164–71. [10] Dohle H, Divisek J, Mergel J, Oetjen HF, Zingler C, Stolten D. Recent developments of the measurement of the methanol permeation in a direct methanol fuel cell. J Power Sources 2002;105:274–82. [11] Arico AS, Srinivasan S, Antonucci V. DMFCs: From fundamental aspects to technology development. Fuel Cells 2001;1:133–61. [12] Yang T. Composite membrane of sulfonated poly(ether ether ketone) and sulfated poly(vinyl alcohol) for use in direct methanol fuel cells. J Membr Sci 2009;35:2160–75. [13] Chiu Y-J, Lien H-C. A strategy of estimating fuel concentration in a direct liquid-feed fuel cell system. J Power Sources 2006;159:1162–8. [14] Zhao H, Shen J, Zhang J, Wang H, Wilkinson DP, Gu CE. Liquid methanol concentration sensors for direct methanol fuel cells. J Power Sources 2006;159:626–36. [15] Ha TJ, Kim J-H, Joh H-I, Kim S-K, Moon G-Y, Lim T-H, et al. Sensor-less control of methanol concentration based on estimation of methanol consumption rates for direct methanol fuel cells systems. Int J Hydrogen Energy 2008;33:7163–71. [16] Shen K-S, Wan C-C, Wang Y-Y, Yu TL, Chiu Y-J. An algorithm for sensor-less fuel control of direct methanol fuel cells. J Power Sources 2010;195:4785–95. [17] Chang CL, Chen CY, Sung CC, Liou DH. Fuel sensor-less control of a liquid feed fuel cell system under steady load for portable applications. J Power Sources 2007;164:606–13. [18] Chang CL, Chen CY, Sung CC, Liou DH. Fuel sensor-less control of a liquid feed fuel cell under dynamic loading conditions for portable power sources (I). J Power Sources 2008;182:133–40. [19] Chang CL, Chen CY, Sung CC, Liou DH, Chang CY, Cha HC. Fuel sensor-less control of a liquid feed fuel cell under dynamic loading conditions for portable power sources (II). J Power Sources 2010;195:1427–34. [20] Chen CY, Liu DH, Huang CL, Chang CL. Portable DMFC system with methanol sensor-less control. J Power Sources 2007;167:442–9. [21] Chen CY, Chang CL, Sung CC. Operation characteristic analysis of a direct methanol fuel cell system using the methanol sensor-less control method. Fuel Cells 2012;12:883–91. [22] Arisetty S, Jacob CA, Prasad AK, Advani SG. Regulating methanol feed concentration in direct methanol fuel cells using feedback from voltage measurements. J Power Sources 2009;187:415–21. [23] Lian K-Y, Yang C-M. Sensor-less adaptive fuel concentration control for direct methanol fuel cells under varying load. J Power Sources 2013;231:239–45. [24] An M-G, Mehmood A, Ha HY. (2014) A sensor-less methanol concentration control system based on the feedback of stack temperature, Appl Energy. [25] Skogestad S. Simple analytic rules for model reduction and PID controller tuning. J Process Control 2003;13:291–309. [26] Ziegler JG, Nichols NB. Optimum settings for automatic controllers. Trans ASME 1942;64:759–68.