Annex 1. Field Report: Solar Electric Light Fund Energy Harvest Control Study

Size: px

Start display at page:

Download "Annex 1. Field Report: Solar Electric Light Fund Energy Harvest Control Study"

David Knight
5 years ago
Views:

1 Annex 1. Field Report: Solar Electric Light Fund Energy Harvest Control Study

2 3 Procedures: 3.1 Incoming Inspection and Labeling: The components were unpacked and labeled according to CFV Solar convention. The CFV IDs and the component descriptions were recorded in the Excel document. 3.2 Outdoor Energy Monitoring: Two rounds of monitoring were completed at CFV s outdoor test yard. Figure 3.2 shows an aerial view of the test yard with the location of the three 360 Watt PV solar arrays and the electronics shed. The solar modules were mounted side-side on a fixed tilt rack and connected four in parallel for each of the three systems. The DC cables were run above ground to the electronics shed containing all the vaccine refrigerators, monitoring equipment, batteries, charge controllers, and custom controllers. The shed also held two 6 kw inverters and some monitoring equipment related to a separate project. The resistive loads for Round 1 were mounted in metal boxes outside the shed. Round 2 of testing replaced the resistor load with commercially available DC lights. Current, voltage, and temperatures were all monitored at points within the system and recorded in an Agilent data logger. Power and energy were calculated for the PV solar array, the refrigerator, and the load for each system independently. There were two temperature sensors attached to the inside of each vaccine storage compartment, one near the bottom and one near the top of the cooled compartment. Ambient temperature was also measured inside the electronics shed and just outside the shed under the eve. Irradiance data in the plane of the PV solar array was measured and recorded, as well. Once per weekday, the lids were opened and eight water bottles were removed from each fridge and replaced with eight warm water bottles that had were stored in the electronics shed. This procedure was not done on the weekends. The water bottles held 0.5 liters each. The lids were left open for 10 minutes and then closed again until the next day. Figure 3.2 Outdoor test yard at CFV Solar showing locations of PV array and electronics shed Page 2 of 17 A-2

3 The first round of monitoring lasted for four weeks and the second round lasted for one week. Table 3.2 provides the details of each setup. According to the requirements of the system, the energy from the solar arrays was distributed to the vaccine refrigerators as a priority. Excess energy was distributed to the battery and/or the external load, as governed by the logic of the specific custom controller and system configuration under evaluation. Table 3.2 System configurations for Round 1 and Round 2 testing Page 3 of 17 A-3

4 The data was analyzed based on daily averages and based on the individual measurements, depending on the objective. For the daily averages, the data was filtered to provide daily summary statistics based on a balanced number of observations across all systems. The first two weeks of Round 1 data was excluded because of a number of issues. The batteries died within six days for the OR1 and OR2 controllers. After the resistive load was re-configured, the systems generated useful comparative data without draining the batteries. Prior to the batteries depleting, there was an issue with the location of the current shunt for the load monitoring, and one of the shunt monitors was reversed. On June 3 rd, three hours of data are missing between 9:00 am and 12:00 PM, so that day was excluded in the daily summary statistics. On Saturday, June 4th, the data logger lost communication, so June 4 and Sunday, June 5 were excluded from the Round 2 test data. The remaining 15 days from Round 1 and 3 days from Round 2 provided good sample of clean data for the analysis. All three refrigerators continued to run during the setup issues, so the dead batteries affected the energy harvest but not the refrigerator performance. Page 4 of 17 A-4

5 4 Discussion of Results: 4.1 Daily plots: Daily plots were used to monitor the systems on a regular basis. Figure shows trends from a cloudy day for each of three systems being monitored. Figure shows trends for the same three systems from a clear sky day. Each plot consists of a trend for current from the array and current to the fridge and load. The tstat measures voltage change at the thermostat and indicates when the refrigerator is calling for cooling and thus correlates with the current flow to the refrigerator. The upper and lower temperatures are within tolerance (+2ºC to +8ºC) except for the period immediately following the 10 minute period when the lid was opened and bottles were swapped in order to simulate a use case. 5/21/16 fell on a Saturday, so the lids were not opened and thus the temperature profiles were flat. Figures and show similar daily trends for Round 2 testing. The AND1 system was the same for round1 and Round 2 test, but the other two systems were modified. Figure shows daily trends for the PV array current measured from each system and the available current available from the array. The available current was simply calculated as the measured irradiance/1000 * 4 * 5.11, since each array consisted of 4 modules in parallel with an STC current at maximum power of 5.11 A. These plots were included to provide an estimate for the power available from the array that was not being used by the refrigerators, the loads, or the controllers. The same four days were used for all plots. 5/21/16 and 6/8/16 were chosen as the clear sky days for Round 1 and Round 2, respectively. 5/18/16 and 6/7/16 were chosen as the cloudy or partly cloudy sky days for Round 1 and Round 2. Page 5 of 17 A-5

6 Figure Round 1 cloudy day (5/18/16) for AND1, OR1 and OR2 controllers Page 6 of 17 A-6

7 Figure Round 1 clear day (5/21/16) for AND1, OR1 and OR2 controllers Page 7 of 17 A-7

8 Figure Round 2 partly cloudy day (6/7/16) for AND1, AND2 and OR3 controllers Page 8 of 17 A-8

9 Figure Round 2 clear day (6/8/16) for AND1, AND2 and OR3 controllers Page 9 of 17 A-9

OR2 Round 2 AND2 Round 2 OR3 Figure 4.1.

10 Clear Cloudy/Partly Cloudy Round 1 AND1 Round 1 OR1 Round 1 OR2 Round 2 AND2 Round 2 OR3 Figure Actual PV current (blue) versus available current (red) in Amps (0-22 A) versus time of day Page 10 of 17 A-10

4.2 Temperature Trends: Figure 4.2.1 shows the temperature trends from inside the refrigerators for both Round 1 and Round 2 system configurations for the entire monitoring period, both night and day.

11 4.2 Temperature Trends: Figure shows the temperature trends from inside the refrigerators for both Round 1 and Round 2 system configurations for the entire monitoring period, both night and day. The spikes correlate to the lid opening each day. The weekends, holidays, and end of Round 1 show no such spikes because the refrigerators remained closed. The Appliance 2 unit showed the most extreme temperatures in the upper portion of the compartment, peaking around +25ºC following the lid opening. The Appliance 1 and the Appliance 3 peaked around +15ºC. The temperature sensors (TCs) in the Appliance 1 unit were attached directly to the interior wall of the vaccine compartment, while the TCs on the other two units were attached to the wire baskets. Figure Temperature trends inside the refrigerators Page 11 of 17 A-11

12 The majority of the excursions below +2ºC were observed on the AND1 controller system, upper TC. This sensor was deliberately placed inside the vaccine basket but directly in the airflow stream from the ice bank. Figure shows the recovery period for the upper temperature on the AND1 controller following a lid opening event. This example is taken from May 11. The AND1 controller tended to porpoise up and down following a lid closure, and often times exceeded the lower threshold of +2º C. The OR controllers and the AND2 controller tended to recover temperatures in a smooth fashion, thus avoiding the overcorrection. Figure Upper temperature trend inside the AND1/Appliance 3 system May 11, 2106 Page 12 of 17 A-12

4.3 Energy Monitoring: Table 4.3.1 shows minimum, maximum, and average values for the important parameters monitored during this study. The data are presented for Round 1 and Round 2 separately.

13 4.3 Energy Monitoring: Table shows minimum, maximum, and average values for the important parameters monitored during this study. The data are presented for Round 1 and Round 2 separately. The data logger channels were dedicated to each of the refrigerators, so the naming convention in the description field is dual-purposed. For example, in the table the description Appliance 1 OR1/OR3 Upper Temp indicates the fact that the Appliance 1 was controlled with the OR1 controller during Round 1 and the OR3 controller during Round 2. Similarly, the Appliance 2 was controlled by the OR2 in Round 1 and the AND2 in Round 2. The Appliance 3 was controlled by the AND1 in both Round 1 and Round 2. Table Summary statistics for Round 1 and Round 2 data. The temperature data includes all the recorded day time and night time observations. None of the recorded temperatures inside the refrigerator compartments ever dipped below +0.7 C. The irradiance and energy data was summarized to the daily level, and then the maximum, minimum, and average of the daily values were reported for valid days only. The refrigerator energy was summed up for each day and then the minimum, maximum, and average were reported. The Available energy from array (assuming 13% eff.) was calculated from the minimum, maximum, and average of the daily insolation values multiplied by the area of the array (2.6 m 2 ) and then multiplied again by The PV module has a rated efficiency of 13.9% at STC, so the reduced efficiency of 13% is an approximation to account for reduced performance at higher temperatures and different irradiance values. This value was calculated in order to estimate the potential energy available from the array that was not harvested due to lack of load. Finally, the Energy harvested fields represent the energy harvested by the external loads, excluding the refrigerators. In Round 2, the OR3 controller shows a much higher energy harvest than the AND1 or AND2. This could be due to the controller itself or to the load or the combination. The OR3 was dumping load continuously 24 hours per day through the Deka battery, while the AND controllers were dumping load through 8 Pico lamps each on continuously until depleted. Page 13 of 17 A-13

14 Table shows the energy harvest efficiency for each system and round of testing. This metric was calculated by summing up the energy required to run the refrigerator (Fridge Energy), the actual energy harvested across the load (Measured harvested energy), and the estimated available energy from the array assuming a 13% efficiency, and then computing the efficiency per Equation These energy estimates were made for each observation from the 15 complete days in Round 1 testing and the 3 complete days from Round 2. The AND1 controller harvested 20% of the available energy during Round 1, while the OR controllers only harvested 11% and 6%, respectively. The OR3 controller harvested the most energy in Round 2. The energy harvesting was likely limited by the loads connected to the systems. Equation Calculation of the energy harvest efficiency Table Energy harvest efficiency of for each system Page 14 of 17 A-14

Table 4.3.3 shows summary statistics for energy during specific days chosen to reflect a daily snapshot across a range of insolation.

15 Table shows summary statistics for energy during specific days chosen to reflect a daily snapshot across a range of insolation. The potential energy was calculated as a function of the measured insolation, the array area (2.6 m), and an assumed module efficiency of 13%. The energy required to run the refrigerators during Round 1 ranged from 85 Wh to 272 Wh per day during Round 1. The energy consumtion increased to between 249 Wh and 539 Wh during Round 2. The temperature inside the shed was approximately 8ºC hotter during Round 2, which may explain some of the increased energy demand by the refrigerators. The PV array output correlates with insolation for most systems. The PV output is also a function of the total load on the system, so the output can be limited on load side as well as the supply side (available sunlight). The harvested load for the AND1 system dropped dramatically in Round 2 due to a reduced load. In Round 1, AND1 was connected to a battery and a 15 A resistive load. In Round 2, it was connected to 8 Pico lamps, a much reduced load. The load on the OR1/OR3 system changed from a battery and resistive load in Round 1 to a load consisting of a battery and 2 x 12 Watt compact flourescent lights (CFLs) that were constantly illuminated in Round 2. The load on the OR2/AND2 system changed from a battery and resistive load in Round 1 to a reduced load of 8 Pico lamps in Round 2. Finally, the last three columns show the difference between the measured energy out of the PV arry and the combined load from the fridge and the external load. The delta is largest for the AND controlled systems. Based on lab tests done elsewehere on behalf of SELF, the AND controls consume approximately 0.78 A at 18 VDC, which would add up to 337 Wh per day. That s more than enough to account for the additional energy out from the PV array that was not captured by the monitors on the refrigerator loads and the external loads. Based on the data in the table and prior testing, the OR controllers use significantly less energy than the AND controllers. Table Energy statistics for selected days representing a broad range of insolation Page 15 of 17 A-15

$Table 4.3.4 shows the fraction of potential energy harvested by the system. The measured PV array output versus the potential array output ranged from 16% to 63%.$

16 Table shows the fraction of potential energy harvested by the system. The measured PV array output versus the potential array output ranged from 16% to 63%. The largest percentage of available energy from the array was harvested during Round 1 using the AND controller. The energy harvest percentage correlated with the available sunlight during that test period, as well. During a clear sky day with high insolation, the system only harvested 38% of the potential energy. During the cloudy day with low insolation, the system harvested 63% of available energy. This suggests more energy could be harvested on sunny days from this AND system provided there is sufficient load connected to the system. In general, between 37% and 84% of available energy is not being harvested due to the limited load and control logic. Table Energy harvesting efficiency Page 16 of 17 A-16

17 5 Revision Control Revision Date Description of Change Original June 17, 2016 Original July 14, 2016 Corrected upper temps for Appliance 1 in Round 2. Minor edits. Expanded the Daily Plot section August 19, 2016 Replaced the specific manufacturer names with Appliance 1, 2, and 3 Page 17 of 17 A-17

Designing Stand Alone Systems. Overview, components and function, Elements in Design

Designing Stand Alone Systems. Overview, components and function, Elements in Design Designing Stand Alone Systems Overview, components and function, Elements in Design What Stand Alone System Does Loads that are Reasonable for a Stand Alone System to Power: Yes or No Dishwasher? Refrigerator