Case Study. R510 Residential Energy Storage Newington Smart Home Trial. October - December 2012

Transcription

1 Case Study R510 Residential Energy Storage Newington Smart Home Trial October - December February 2012

2 Background As part of the larger Smart Grid, Smart Cities (SGSC) project, Ausgrid is conducting a Smart Home trial in Newington, Sydney, involving the testing of new technologies that will define the house of a greener and more resource-conscious future. This trial, which began in July 2010, involves the practical experiment of placing a typical Australian family in a typical Australian suburb for one year in a house that is fitted with the latest energy and water-saving devices. The family also has access to an electric vehicle (EV). The house uses a CFCL Blue-Gen fuel cell and solar PV panels for energy generation, with a connection to the grid when these are not sufficient. RedFlow s energy storage system (ESS) provides an additional supply during peak times. Fundamentally, the aims of RedFlow s products include: Demonstrating that the house can operate on net-zero intake from, or net-export to, the grid Demonstrating that the house is able to operate without drawing from the grid Reducing peak demand from the grid and shift this to off-peak times Managing key loads and energy generators to achieve other aims o e.g. Avoiding charging the EV during peak times RedFlow has installed two products for this trial: the ESS and the energy management system (EMS). The ESS utilises an inverter and a Remote Terminal Unit (RTU), which stream data continuously to RedFlow s HOST in Brisbane. The EMS controls parameters of system operation such as energy flows and previously, EV charging. Initially in July 2010, a 10kWh lead acid (LA) ESS was deployed to the site. In September 2010, a 10kWh zinc bromine module (ZBM) was added with a new battery management system (BMS) and new RTU firmware to create a hybrid ESS that augmented the existing LA battery bank. The ZBM was taken out of service due to an operational failure in May 2011 (see Appendix C ZBM Fault Report) and the Smart Home once again returned to operation using purely LA storage. In October 2011, an R510 was installed in the Smart Home to replace the previous ESS. RedFlow has been monitoring many parameters of the Smart Home s energy system since the beginning of the trial. Based on early feedback, alterations have been made regarding the ways in which the energy storage system reacts to the Smart Home s energy demand, and these have contributed to the design of the hardware and firmware of the R510. ii

3 Executive Summary As part of the larger Smart Grid, Smart Cities (SGSC) project, Ausgrid is conducting a Smart Home trial in Newington, Sydney, involving the testing of new technologies that will define the house of a greener and more resource-conscious future. This trial, which began in July 2010, involves the practical experiment of placing a typical Australian family in a typical Australian suburb for one year in a house that is fitted with the latest energy and water-saving devices. RedFlow has provided a series of energy storage systems for the project, which has been continually improved as the project progresses. The most recent model of energy storage system installed as the Smart Home is the zinc bromide-based R510. The operational results and benefits of the R510 in this application were analysed, and the lessons learnt presented. This produced the following key conclusions: The appropriateness of this size and rating of energy storage to the needs of the Smart Home household load. The significant effect that charging the EV has on Smart Home load, often requiring the import of power from the grid. The need for a fuel cell or other similar reliable form of embedded generation to back-up solar generation (which is often insufficient to power the Smart Home load throughout the day, even with the use of energy storage). The effectiveness of the R510 in conjunction with embedded generation in greatly reducing grid import to below 8% of the time. iii

4 Contents Background... ii Executive Summary... iii Contents... iv List of Figures... v Summary Operation ESS and EMS Solar Fuel Cell Load Subsystem Integration Remote Monitoring Results ESS and EMS Solar Fuel Cell Load Lessons Learnt ESS and EMS Solar Fuel Cell Load Conclusions Appendix A List of Abbreviations Appendix B Smart Home Installation Drawing Appendix C ZBM Fault Report iv

5 List of Figures FIGURE 1: POWER FLOWS IN THE SMART HOME... 1 FIGURE 2: A MAP SHOWING THE LOCATION OF THE NEWINGTON SMART HOME... 2 FIGURE 3: R FIGURE 4: SCHOTT THIN FILM SOLAR PANEL... 4 FIGURE 5: BLUE GEN FUEL CELL... 5 FIGURE 6: A CONSTANT 1.4KW BLUE GEN OUTPUT CURVE... 5 FIGURE 7: A VARYING BLUE GEN OUTPUT CURVE TO MATCH THE SMART HOME'S TYPICAL CONSUMPTION... 5 FIGURE 8: A TYPICAL SMART HOME LOAD OVER THE PERIOD OF ONE DAY (27 OCTOBER 2011 SHOWN)... 6 FIGURE 9: THE COMMUNICATIONS LINK BETWEEN THE NEWINGTON SMART HOME AND MONITORING AT REDFLOW OFFICES... 7 FIGURE 10: A MICROSOFT EXCEL-GENERATED GRAPH SHOWING THE PERFORMANCE OF THE SMART HOME S ENERGY SYSTEM ON 10 DECEMBER FIGURE 11: THE EFFECTS OF USING EMBEDDED GENERATION AND AN ESS WITH THE SMART HOME (30 OCTOBER 2011 SHOWN)... 8 FIGURE 12: THE PERFORMANCE OF THE ESS IN LIMITING IMPORT OF POWER FROM THE GRID... 9 FIGURE 13: TIME SPENT EXPORTING TO, IMPORTING FROM OR OPERATING AT ZERO WITH THE GRID. PEAK PERIODS ARE DEFINED FROM 4PM TO 8PM EACH DAY FIGURE 14: PERFORMANCE OF THE SOLAR PANELS, AND THE EFFECT OF SOLAR EXPOSURE FIGURE 15: THE INTERMITTENCY OF SOLAR GENERATION OVER ONE DAY FIGURE 16: ENERGY GENERATED EACH DAY BY THE FUEL CELL FIGURE 17: PERFORMANCE OF THE ESS DURING A FUEL CELL OUTAGE ON 22 OCTOBER FIGURE 18: HOUSEHOLD AND EV LOADS USED IN THE SMART HOME PER DAY FIGURE 19: THE RELATIONSHIP BETWEEN CHARGING THE EV AND THE NEED TO IMPORT POWER FROM THE GRID v

6 Summary This project has shown that it is technically feasible to fulfill the four main aims noted in the Background Section for such a residential house. It has also shown, however, that there are obstacles to achieving those goals when large continuous loads or an inappropriate BMS are used. This Case Study provides an overview of the Newington Smart Home project, and how the RedFlow ZBM and R510 ESS functioned in their roles of providing energy storage to successfully attain the project goals. While focusing on the period of R510 operation between October and December 2011, it will also draw on lessons learnt from other times during the Smart Home project. The household was provided with a 1kW rooftop solar panel, a 0.5kW solar film on the pergola roof, as well as a 1.5kW (gas) fuel cell for electricity generation. The Smart Home s load consisted of household appliances including a fridge, washing machine, dryer, computer and entertainment system. The EV was also charged through a specially-installed power point in the garage. The connections between these elements, as well as the R510, are shown below in Figure 1. Grid Solar Panels House Load Fuel Cell R510 ESS Direction of Power Flow EV Load Figure 1: Power flows in the Smart Home The Smart Home is located in the western suburbs of Sydney, Australia. This is shown in the map below in Figure 2: 1

7 Figure 2: A map showing the location of the Newington Smart Home This Case Study will be divided into three main sections: Operation, Results and Lessons Learnt. Each section will address the significance of the generation, load and energy storage subsystems. Appendix A is a list of abbreviations used in this Case Study. Appendix B is an electrical installation drawing of the Smart Home. Appendix C provides a full fault report for the ZBM fault that occurred in May

8 1 Operation 1.1 ESS and EMS The R510 ESS (see Figure 3) installed at the Newington Smart Home comprises many elements, including the ZBM, battery controller (BC), SMA Sunny Backup Inverter, RTU and 3G modem. Some of these elements are shown in the installation drawing in Appendix B Smart Home Installation Drawing. Figure 3: R510 The ESS provides energy storage to the Smart Home, while the EMS controls power flows from generation, to the loads, and to and from the ZBM and grid. It does this through the use of current transformers (CTs) that measure the output of generation, as well as the consumption of loads. Grid import and export are measured internally by the SMA inverter. The main aim of the EMS operation is to achieve zero import/export from the grid for as long as possible. This control of power flow is set to operate under the following order of preferences: 1. The household and EV loads must be satisfied under all circumstances. 2. This power will first come from the generation at the Smart Home (i.e. the solar panels and the fuel cell). 3. If this generation is not sufficient, the ZBM will discharge at the required power output (maximum 5kW through inverter) until it is completely depleted (i.e. 0% state of charge (SOC)). If this generation is larger in magnitude than the Smart Home s consumption, the surplus power will be used to charge the ZBM until it reaches its nominated maximum SOC (this was set to 50% for much of the time as there was rarely enough surplus power to reach this SOC, and setting the SOC lower than 100% reduces stress on the internals of the ZBM). 3

9 4. If generation is not sufficient to power the Smart Home consumption and the ZBM is completely depleted, the system will import power from the grid to make up the difference. If the generation is larger in magnitude than the Smart Home s consumption and the ZBM has reached its maximum set SOC, the surplus power from the generation is exported back to the grid. NOTE: 1. The ZBM is only charged using surplus power from generation (fuel cell or solar panels) and not from the grid. This is done to achieve the goal of zero import from the grid. 2. The ZBM, according to operational requirements, was stripped every night. In order to do this, the ZBM was fully discharged after 11:30pm, followed by a strip. This meant that there was no energy storage available for use in the Smart Home for about an hour very early each morning during the strip cycle. This was timed to coincide with times of lowest Smart Home loads (as well as after the EV usually finished charging). 3. The EMS could not control the outputs of the generation. That is, the fuel cell operated on an automatic output curve and the solar generation could not be disconnected if extra generation was no longer needed. 1.2 Solar There are two solar panels installed in the Smart Home. One is a BP 1kW PV, situated on the roof of the house. The other doubles as a sun shade on the roof of the pergola area, and is a 500W Schott thin film solar panel (see Figure 4). Together at times of the highest solar exposure, the solar panels can generate up to 1.5kW. However, this only occurs for a short period of time in the middle of the day during the summer months. For the majority of the time, the solar panels generate a fraction of their 1.5kW peak. The solar panels have operated without any significant problems over the duration of the trial. They are also a supplementary source of electricity to the Blue Gen fuel cell described below. Figure 4: Schott thin film solar panel 4

10 12:00 AM 1:30 AM 3:00 AM 4:30 AM 6:00 AM 7:30 AM 9:00 AM 10:30 AM 12:00 PM 1:30 PM 3:00 PM 4:30 PM 6:00 PM 7:30 PM 9:00 PM 10:30 PM 12:00 AM 1:30 AM 3:00 AM 4:30 AM 6:00 AM 7:30 AM 9:00 AM 10:30 AM 12:00 PM 1:30 PM 3:00 PM 4:30 PM 6:00 PM 7:30 PM 9:00 PM 10:30 PM kw 1.3 Fuel Cell The CFCL Blue Gen fuel cell (see Figure 5), operated most efficiently at 1.4kW output, uses gas to produce a daily cycle of generation that is the primary source of electricity generation for the house. The fuel cell was not capable of dynamically reacting to the load and hence, over the duration of the trial, several daily output curves were used to balance generation with consumption. Figure 5: Blue Gen fuel cell The different output curves in use between the period of October and December 2011 are shown below in Figure 6 and Figure Blue Gen Output Curve October - 28 November 2011 Blue Gen Output Curve 29 November - December Figure 6: A constant 1.4kW Blue Gen output curve Figure 7: A varying Blue Gen output curve to match the Smart Home's typical consumption 1.4 Load The load profile of the Smart Home is similar to that of a typical Australian household. Although the load profile varies, a graph of a typical day s load is shown in Figure 8. There is a base load of approximately 300W, with usual peaks in the morning (6-10am) and evening (5-9pm). These peaks can reach up to approximately 5kW for short periods of time. 5

11 12:00 AM 1:00 AM 2:00 AM 3:00 AM 4:00 AM 5:00 AM 6:00 AM 7:00 AM 8:00 AM 9:00 AM 10:00 AM 11:00 AM 12:00 PM 1:00 PM 2:00 PM 3:00 PM 4:00 PM 5:00 PM 6:00 PM 7:00 PM 8:00 PM 9:00 PM 10:00 PM 11:00 PM kw Throughout the trial period, a Mitsubishi i-miev was made available to the Smart Home family to use. This is a fully electric vehicle with a 16kWh capacity lithium-ion battery that charges at a constant rate of approximately 3kW. As can be seen in Figure 8, the Smart Home s energy consumption increases significantly when the EV is charging. As such, the EV often requires more power than the household at peak periods. The EV was set to charge only between 9pm and 6am each night by an externally-controlled power point. There were, however, some issues with set times throughout the trial Typical Smart Home Load Time of Day Total Usage EV Usage House Usage Figure 8: A typical Smart Home load over the period of one day (27 October 2011 shown) 1.5 Subsystem Integration The R510 ESS operates as designed when used with all other subsystems in the Smart Home electricity system. It successfully stores surplus energy from the solar panels as well as from the fuel cell for times when consumption is greater than generation, thereby limiting the need to import power from the grid. However, large continuous loads often necessitate importing power from the grid. During the period of October to December 2011, the EV was the single load which required power from the grid, as the ZBM could not provide the required energy. This is because the EV s capacity is 16kWh (though the EV was charged to about 10kWh on most occasions since the battery was not entirely depleted), and this capacity was on many occasions larger than the energy stored in the ZBM that day. This problem was exacerbated by the fact that the EV was set to charge directly after the hours of peak demand, which often used energy from the ZBM as generation from the fuel cell was not sufficient. Therefore, had the Smart Home family not utilized the EV, the project would have been much more successful in attaining its zero import from the grid goal. The evaporative cooling system was another appliance that caused the system to import power from the grid in months before the R510 was installed. This cooling system was installed instead of a conventional air conditioning system, as it was claimed to be more energy efficient. However, there were two systems installed, each consuming a constant approximate 1.3kW each when operating. As such, when both systems were in use, the constant 2.6kW load was too large for the fuel cell and ESS supplement power to supply for long periods. 6

12 1.6 Remote Monitoring In order to ensure that the R510 was operating as expected, as well as to change any parameters of the system, there was a 3G modem in the R510 unit allowing for remote communications to the RedFlow offices in Brisbane (see Figure 9). Newington Smart Home ESS Brisbane Figure 9: The communications link between the Newington Smart Home and monitoring at RedFlow offices All parameters for the system are sent regularly through to the HOST program in Brisbane. This allows the team at RedFlow to monitor the operations at the Smart Home, and make adjustments as required without the need for trips to Sydney. The DataViewer software is also used to create graphs and.csv files for further analysis. The vast majority of changes can be made remotely through the HOST. However, if any RTU firmware, BC firmware or any hardware change is needed, this must be done on site in Sydney. The.csv files can be used in Microsoft Excel to further analyse data, and create more complex graphs, like the one presented below in Figure 10. This shows the generation at the Smart Home, as well as the household and EV loads. It also shows how much power was imported and exported to the grid, and how the ZBM charged and discharged depending on the generation and consumption at the Smart Home. Figure 10: A Microsoft Excel-generated graph showing the performance of the Smart Home s energy system on 10 December

13 12:00 AM 1:00 AM 2:00 AM 3:00 AM 4:00 AM 5:00 AM 6:00 AM 7:00 AM 8:00 AM 9:00 AM 10:00 AM 11:00 AM 12:00 PM 1:00 PM 2:00 PM 3:00 PM 4:00 PM 5:00 PM 6:00 PM 7:00 PM 8:00 PM 9:00 PM 10:00 PM 11:00 PM kw 2.1 ESS and EMS 2 Results 50.56% Average R510 Efficiency kWh Total Energy Exported to Grid 97.23kWh Total Energy Imported from Grid The following graph (see Figure 11) shows the effect that using an R510 ESS, in conjunction with embedded generation, can have on reducing the import of power from the grid to zero. It can be seen that when embedded generation is used in isolation without an ESS, peaks still occur when consumption is greater than generation. As is the case with the 30 October 2011 below, these peaks often occur during periods of peak demand, which is the time when it is of most benefit to utilities to limit consumption seen by the grid. However, by using an R510, it can be seen that importing power from the grid can be completely avoided. Comparison of Grid Import with and without Embedded Generation and ESS Consumption becomes greater than generation and grid power is needed. An ESS negates the need for grid import Smart Home Consumption (Grid Import without Embedded Generation and ESS) Grid Import with Embedded Generation but without ESS Grid Import with Embedded Generation and ESS Figure 11: The effects of using embedded generation and an ESS with the Smart Home (30 October 2011 shown) The graph below (see Figure 12) shows how the ESS limited the system from importing energy from the grid over the period of R510 testing. As can be seen, there was very little energy imported from the grid. Where significant energy has been imported, explanations have been given on the graph. Reasons include a fuel cell outage (greatly reducing the Smart Home s generation, therefore requiring grid import), unusually high load usage (see Section 2.4) and days when the health of the ZBM was evaluated using calibration tests. The effect of charging the EV on grid import performance is outlined in greater detail in Section 2.4. Grid export has also been shown on the graph, along with net grid import over the course of a 24 hour calendar day. This shows that for that vast majority of days, the EMS exported far more power to the grid than it imported. 8

14 % 11-Oct 13-Oct 15-Oct 17-Oct 19-Oct 21-Oct 23-Oct 25-Oct 27-Oct 29-Oct 31-Oct 2-Nov 4-Nov 6-Nov 8-Nov 10-Nov 12-Nov 14-Nov 16-Nov 18-Nov 20-Nov 22-Nov 24-Nov 26-Nov 28-Nov 30-Nov 2-Dec 4-Dec 6-Dec 8-Dec 10-Dec 12-Dec 14-Dec 16-Dec 18-Dec 20-Dec kwh Daily Import from and Export to the Grid EV charged Fuel Cell Outage EV Charged /High Load Use ZBM Calibration Test Days Total Energy Exported (kwh) Total Energy Imported (kwh) Net Grid Import over Day Figure 12: The performance of the ESS in limiting import of power from the grid The EMS can at any time be operating in three grid modes: importing power from the grid, exporting power to the grid, or holding zero import/export. The percentage of time spent in each grid mode in terms of peak and off-peak periods is shown in Figure 13. It can be seen that the system successfully avoided importing power from the grid for over 92% of the trial period Time Spent in Grid Modes Export Import Zero Off-Peak Peak Figure 13: Time spent exporting to, importing from or operating at zero with the grid. Peak periods are defined from 4pm to 8pm each day. In terms of utilities interests, it is of most value for the Smart Home to export power to the grid during peak periods and import power during off-peak periods. This relieves the stress on overloaded infrastructure during peak periods, and helps limit voltage rise during off-peak periods, especially when solar feed-in from other sources exacerbates this problem. Energy storage is effectively the only device in the Smart Home system that can realize this for the utility since the Smart Home load is reflective of typical household loads in similar areas. As can be seen in Figure 12, there were two calibration tests undertaken for the ZBM installed in the R510 at the Newington Smart Home. This is good practice, and should be performed regularly to any in-service ZBM to gauge its health and efficiency accurately. At the time of the R510 trial at the Smart Home, ZBM calibration tests were not standard. It was regular monitoring of the system, and regular calculations of ESS efficiency that alerted RedFlow to the possible ill health of the ZBM. This then prompted RedFlow to confirm this through calibration tests. The health of the ZBM was then assessed. It was concluded that due to manufacturing faults, which had been rectified by the time this fault was identified at the Smart Home, the ZBM should no longer be in service as there was a heightened chance of leaks and other operational issues. 9

15 12:00 AM 12:45 AM 1:30 AM 2:15 AM 3:00 AM 3:45 AM 4:30 AM 5:15 AM 6:00 AM 6:45 AM 7:30 AM 8:15 AM 9:00 AM 9:45 AM 10:30 AM 11:15 AM 12:00 PM 12:45 PM 1:30 PM 2:15 PM 3:00 PM 3:45 PM 4:30 PM 5:15 PM 6:00 PM 6:45 PM 7:30 PM 8:15 PM 9:00 PM 9:45 PM 10:30 PM 11:15 PM kw 11-Oct 13-Oct 15-Oct 17-Oct 19-Oct 21-Oct 23-Oct 25-Oct 27-Oct 29-Oct 31-Oct 2-Nov 4-Nov 6-Nov 8-Nov 10-Nov 12-Nov 14-Nov 16-Nov 18-Nov 20-Nov 22-Nov 24-Nov 26-Nov 28-Nov 30-Nov 2-Dec 4-Dec 6-Dec 8-Dec 10-Dec 12-Dec 14-Dec 16-Dec 18-Dec 20-Dec Generation (kwh) Solar Exposure (MJ/m2) 2.2 Solar kWh Total Energy Generated by Solar Panels 3.31kWh Average Daily Energy Generation by Solar Panels The daily performance of the solar panels is shown in Figure 14 below. Also shown is the solar exposure of each day in the Newington area (this information was taken from the Australian Bureau of Meteorology please note that solar exposure data for 14 December was unavailable). There is a clear correlation between the two sets of data. It is clear that even though the Smart Home sometimes sources over 6kWh of energy from its solar panels, it cannot rely on a set amount of energy on a given day as there is a great deal of variation due to the intermittent nature of solar power, even on a daily basis. Daily Solar PV Generation Total Solar Generation (kwh) Solar Exposure (MJ/m2) Figure 14: Performance of the solar panels, and the effect of solar exposure The intermittent nature of solar power over just one day can be seen in Figure 15. The solar curve for 28 November shows a relatively smooth curve to be expected from a sunny day. However, the curve for 29 November shows a great deal of intermittency. While it can be seen from Figure 14 that the 29 November received slightly more solar exposure, the effect of clouds passing over the path from the sun to the solar panels at the Smart Home resulted in almost identical daily totals of solar generation for both 28 and 29 November Position of solar panels greatly increases solar exposure at approx. 9am Solar Generation Over One Day Small clouds passing overhead Large cloud passing overhead 28 November 29 November Figure 15: The intermittency of solar generation over one day 10

16 12:00 AM 1:00 AM 2:00 AM 3:00 AM 4:00 AM 5:00 AM 6:00 AM 7:00 AM 8:00 AM 9:00 AM 10:00 AM 11:00 AM 12:00 PM 1:00 PM 2:00 PM 3:00 PM 4:00 PM 5:00 PM 6:00 PM 7:00 PM 8:00 PM 9:00 PM 10:00 PM 11:00 PM kw 11-Oct 13-Oct 15-Oct 17-Oct 19-Oct 21-Oct 23-Oct 25-Oct 27-Oct 29-Oct 31-Oct 2-Nov 4-Nov 6-Nov 8-Nov 10-Nov 12-Nov 14-Nov 16-Nov 18-Nov 20-Nov 22-Nov 24-Nov 26-Nov 28-Nov 30-Nov 2-Dec 4-Dec 6-Dec 8-Dec 10-Dec 12-Dec 14-Dec 16-Dec 18-Dec 20-Dec kwh 2.3 Fuel Cell kWh Total Energy Generated by Fuel Cell 29.27kWh Average Daily Energy Generation by Fuel Cell The graph below shows the total energy generation per day for the fuel cell (see Figure 16). It can be clearly seen where there was an outage for approximately 36 hours over 25 and 26 November. It can also be seen that the daily output of the fuel cell significantly reduced when the output curve was altered. Daily Fuel Cell Generation Output curve change 25 Fuel cell outage Figure 16: Energy generated each day by the fuel cell Over the course of the trial, the fuel cell experienced several long outages. This put extra pressure on the energy storage to perform, while also requiring extra power from the grid to power the Smart Home load. The fuel cell also experienced many momentary outages, which the energy storage compensated for by quickly reacting to discharge (see Figure 17). As such, the load was always satisfied Blue Gen Fuel Cell Outage Performance Fuel cell outage ZBM discharges to satisfy load ZBM discharges in preparation to strip InvPwrAt [kw] Blue Gen Power Total Usage Figure 17: Performance of the ESS during a fuel cell outage on 22 October 11

17 11-Oct 13-Oct 15-Oct 17-Oct 19-Oct 21-Oct 23-Oct 25-Oct 27-Oct 29-Oct 31-Oct 2-Nov 4-Nov 6-Nov 8-Nov 10-Nov 12-Nov 14-Nov 16-Nov 18-Nov 20-Nov 22-Nov 24-Nov 26-Nov 28-Nov 30-Nov 2-Dec 4-Dec 6-Dec 8-Dec 10-Dec 12-Dec 14-Dec 16-Dec 18-Dec 20-Dec kwh 11-Oct 13-Oct 15-Oct 17-Oct 19-Oct 21-Oct 23-Oct 25-Oct 27-Oct 29-Oct 31-Oct 2-Nov 4-Nov 6-Nov 8-Nov 10-Nov 12-Nov 14-Nov 16-Nov 18-Nov 20-Nov 22-Nov 24-Nov 26-Nov 28-Nov 30-Nov 2-Dec 4-Dec 6-Dec 8-Dec 10-Dec 12-Dec 14-Dec 16-Dec 18-Dec 20-Dec kwh 2.4 Load kWh Total Smart Home Load Used kWh Total Household Load Used kWh Total EV Load Used 15.39kWh Average Daily Energy Use by Smart Home 13.15kWh Average Daily Energy Used by Household 5.48kWh Average Daily Energy Used by EV (on days of charging) The daily load consumption of the Smart Home is given below in Figure 18, showing household and EV components. It can be seen that the Smart Home s household load stays relatively constant, while charging the EV can increase the overall load usage significantly. The resident family did not have access to air-conditioning or any other high-consumption appliance for cooling during the R510 s period of operation. As such, household consumption did not rise significantly with temperature toward the end of the period Daily Load Consumption Resident family on holiday Household Load Usage (kwh) EV Load Usage (kwh) Figure 18: Household and EV loads used in the Smart Home per day The effect of a large continuous load, such as the EV, is significant. This is illustrated in the graph below (see Figure 19). It can be seen that there is a clear relationship between EV charging and large grid energy import. This signifies that the R510 is not suited to such a large load, and would need design modifications if it were to be more successful in the goal of avoiding importing power from the grid while also satisfying an EV load EV Charging Consumption and Import from the Grid Fuel cell outage ZBM Calibration Tests EV Load Usage (kwh) Total Energy Imported (kwh) Figure 19: The relationship between charging the EV and the need to import power from the grid 12

18 3 Lessons Learnt 3.1 ESS and EMS The R510 at the Smart Home was largely successful at avoiding the import of power from the grid. However, there are several important lessons that can be learnt from this trial that can be adapted to future use of energy storage in this type of application: The ESS and EMS can successfully take inputs from embedded generation and control power flows into the load to limit power import from the grid. The Smart Home never experienced a power outage over this R510 trial period. Independently, the Smart Home s household load size and profile are appropriate for the 10kWh of storage in the R510. On the vast majority of occasions, the ESS and EMS have prevented the need to import power from the grid for the household load. Prior knowledge of whether the EV is going to be charged on a particular night would be highly beneficial (though difficult to obtain reliably). This could trigger one of two discharge/strip modes in the EMS. The first is if the EV is going to be charged, the EMS will only supply to the Smart Home the extra power it needs during peak periods and conserve the rest of the stored energy in the ZBM for aiding the embedded generation in charging the EV. If the EV were not going to be charged, the second mode would be initiated whereby extra energy stored in the ZBM could be exported to the grid during peak periods to help relieve network stresses. 3.2 Solar The solar panels produced a great deal less electricity than the fuel cell over this trial, even though they are both rated at 1.5kW. This is due to the intermittent nature of solar energy, and the fact that it can only produce electricity when the sun is shining. The following lessons were learnt from using the solar panels at the Smart Home: The solar panels were not sized large enough for the Smart Home to source a significant portion of its electricity from renewable sources. Instead, the majority came from the fuel cell, which consumes gas, a fossil fuel. A larger size of solar panel would in these respects have been more beneficial. The period of R510 testing in the Smart Home trial was in the later spring and early summer months. The output of the solar panels is markedly lower in winter than the results presented for October to December. Therefore, the Smart Home has an even greater reliance on the fuel cell during winter. 3.3 Fuel Cell The fuel cell provided the primary source of electricity for the Smart Home. The following lessons can be drawn from the use of the fuel cell, and its two output curves used in the R510 trial: The second output curve that loosely followed the load profile was more effective at meeting the goals of the Smart Home. It reduced the energy produced by the fuel cell that was exported straight to the grid, often during off-peak periods when this extra energy is not of use to utilities. 13

19 A constant source of energy such as the fuel cell is required for embedded generation in conjunction with this size of energy storage to effectively limit power import from the grid, unless the capacity and generation of intermittent renewable sources is greatly increased, or can to some degree be guaranteed. This is especially true with the use of the EV that greatly increases the daily load of the Smart Home on days that cannot currently be predicted. The solar panels were not sized large enough for the Smart Home to source a significant portion of its electricity from renewable sources. Instead, the majority came from the fuel cell, which consumes gas, a fossil fuel. A larger size of solar panel would in these respects have been more beneficial. The period of R510 testing in the Smart Home trial was in the later spring and early summer months. The output of the solar panels is markedly lower in winter than the results presented for October to December. Therefore, the Smart Home has an even greater reliance on the fuel cell during winter. 3.4 Load The household loads used in the Smart Home produce consumption curves reflective of most residential buildings in Australia. As such, there is a usual morning and evening peak, with consumption of base load throughout the rest of the day. The use of embedded generation and energy storage, however, manipulates the consumption curve seen by the grid for the Smart Home. The following lessons can be drawn from experiences with loads: The EV is a large and continuous load. Its capacity of 16kWh is far greater than the capacity of the 10kWh of storage in the R510 installed at the Smart Home. Charging the EV thus often required importing power from the grid, as EV charging times were set to be directly after peak periods and hence, the storage was already partially discharged. It was also unknown to RedFlow when the EV would be charged this was decided by the Smart Home resident family. Therefore, RedFlow could not plan to conserve its storage for EV charging if it did this every day, the stored energy would be redundant on most days and would have been more effectively utilized if it had been exported to the grid during peak periods. Charging the EV everyday would have allowed the system to more effectively avoid importing power from the grid. Instead, the resident family charged the EV only every few days, and usually when the battery was significantly discharged. This meant that the energy storage was required to supplement the embedded generation for a longer period of time, increasing the chance that it would completely discharge and require grid power to charge the EV. The resident family was informed of this and while they did charge the EV more frequently towards the end of the trial, it could still have been improved. Communication with electricity consumers is needed in such instances as the Smart Home trial to effectively relay to them the ways they should use electricity to most effectively utilize the capabilities of energy storage and embedded generation. Independently, the Smart Home s household load size and profile are appropriate for the 10kWh of storage in the R510. On the vast majority of occasions, the ESS and EMS have prevented the need to import power from the grid for the household load. The correct operation of the EV charge controller is imperative to avoiding importing power from the grid during peak periods. On some occasions, the charge controller incorrectly allowed the EV to charge during peak periods, exacerbating the problems of importing power from the grid. 14

20 4 Conclusions Overall, the Smart Home R510 trial has shown that it is technically feasible to use this ESS in integrating grid-connected embedded generation to power a typical household load. It has shown that this can be done using minimal power imported from the grid, most of which was used to help charge the EV when the ZBM was depleted. Overall, RedFlow has learnt many valuable lessons over the course of the Smart Home trial, and will use these to improve upon their technology for future similar applications. These have included: The appropriateness of this size and rating of energy storage to the needs of the Smart Home household load. The significant effect that charging the EV has on Smart Home load, often requiring the import of power from the grid. The need for a fuel cell or other similar reliable form of embedded generation to back-up solar generation (which is often insufficient to power the Smart Home load throughout the day, even with the use of energy storage). The effectiveness of the R510 in conjunction with embedded generation in greatly reducing grid import to below 8% of the time. 15

21 Appendix A List of Abbreviations BC - Battery Controller BMS - Battery Management System CT - Current Transformer EMS - Energy Management System ESS - Energy Storage System EV - Electric Vehicle LA - Lead Acid PV - Photovoltaic RTU - Remote Terminal Unit SGSC - Smart Grid, Smart Cities SOC - State of Charge ZBM - Zinc Bromide Module 16

22 Appendix B Smart Home Installation Drawing 17

23 Appendix C ZBM Fault Report 18

24 Newington Smart Home ZBM Fault on 30 April 2011 Report A D V A N C E D E N E R G Y S T O R A G E S Y S T E M S Newington Smart Home ZBM Fault on 30 April 2011 Report Author: Mio Nakatsuji-Mather Date: 30 June RedFlow Limited. All Rights Reserved. Page 19

25 Newington Smart Home ZBM Fault on 30 April 2011 Report Summary of Fault RedFlow s Energy Storage System (ESS) suffered partial failure on 30 April 2011 when the Zinc Bromine Module (ZBM) battery developed a fault and automatically shut down. The ESS system continues to work normally but with a restricted capacity to store electricity. The cause of the fault was an incorrectly set parameter which resulted in the battery operating outside its design parameters. This resulted in a small internal leak which was fully contained in the battery enclosure, causing no issue to the resident family or the environment. Ausgrid and RedFlow are discussing the best time for a replacement ZBM to be provided. Event Details At approximately 8:10pm on Saturday 30 April 2011, a fault occurred with the ZBM, which is part of RedFlow s ESS. This is installed as part of the Ausgrid Smart Home project implemented in Newington, Sydney. As a result of this fault, there was a small leak of electrolyte through the relief valve of the ZBM. It was contained in the top section of the tank and fully contained within the ZBM enclosure. This leak tripped a leak detector, and the ZBM battery was automatically shut down. As far as RedFlow is aware, the resident family noticed no problems with both the RedFlow ESS and their power delivery. As a result, this fault did not cause any disruption to their daily lives. All other aspects of the ESS and Energy Management System (EMS) continued to operate normally. The house continues to enjoy power as usual, with the lead acid batteries currently assuming the full role of energy storage. After discussions with Ausgrid on Tuesday 3 May (2 May being a public holiday in Queensland), RedFlow sent a team to Newington the following day to assess the problem and remove the leaked electrolyte. It was concluded by the RedFlow team that the battery was irreparable but posed no risk to the house or its residents. Ausgrid have taken this advice and decided to await replacement of the ZBM until the new R510 unit is installed later this year. Safety Issues This fault has demonstrated the inherent safety of the RedFlow ZBM battery design. Electrolyte leaks This fault has demonstrated that the design of RedFlow s ZBM battery is able to contain any internal leaks and shut down the system automatically: The electrolyte leak that occurred as part of the initial fault was well contained inside the ZBM enclosure and was cleaned up during the RedFlow visit on 4 May. The battery at the Newington house is no longer in operation and the ZBM is now in a safe and stable state. There is no risk of any further leaks RedFlow Limited. All Rights Reserved. Page 20

26 Newington Smart Home ZBM Fault on 30 April 2011 Report Bromine odour in the area around the battery During an electrolyte leak there is sometimes a smell similar to that found in a wellchlorinated indoor swimming pool. This is not harmful in the typical concentrations found and dissipates rapidly once the leak is cleaned up: A small bromine gas leak would have occurred as part of the fault on 30 April. As a result, there would have been a hint of bromine in the area surrounding the ZBM unit. However, the resident family did not report any notice of this smell. NOTE: RedFlow is in the process of upgrading its ZBM battery with a zinc powder filter that will eliminate any odour in the event of gas releases through the pressure relief valve. Inherent ZBM battery safety This fault highlights several aspects of the inherent safety of the ZBM battery technology. A ZBM battery is basically a plastic electroplating machine; it does not contain flammable substances such as lithium or sodium and it is highly unlikely that a failure could cause a fire of any sort. Leaks are easily contained, basically requiring no more than a plastic bay for the ZBM electrolyte. Any spills are easily cleaned up with readily available absorbent material, including rags. The area can then be washed down with water. RedFlow response time The final advantage of the ZBM is that a fault does not usually require an immediate on-site response; however, RedFlow accepts that the response time must be within 24 hours for a qualified person to be on site. In this case the response was too slow. RedFlow continues to train qualified persons in each area that it has battery systems deployed. It is also upgrading its customer response system with a new 3 level escalation system. Ausgrid will benefit from both of these improvements. Fault analysis There were several hardware and firmware management issues that culminated in the fault that occurred with this ZBM. The firmware issues were the main cause of the problem. The first firmware issue to occur was on 18 February 2011, when a parameter in the Battery Controller (BC) firmware was changed by RedFlow. This was done to test a slightly modified operating mode for the ZBM battery. In normal operation at Newington, the BC is in Test1 mode, however, on 18 February, this was changed to Manual to assist in resolving an issue of negative battery current at the end of the discharge cycle. This issue was resolved by a BC firmware upgrade on 24 February during a visit by RedFlow to Newington. However, in putting the system into Manual mode, a parameter was automatically changed from a hex value of 0010 to This, in essence, meant that the strip contactor was no longer being connected during the strip phase of the battery charge/discharge cycle (refer to Figure 1). In returning the system to Test1 mode, this issue was not noticed, and the flags hex parameter remained at The strip contactor allows for a quick strip of the battery (essentially, for the battery s voltage to fall to 0V), which is needed on a regular basis to ensure that no zinc is left in the electrode stack RedFlow Limited. All Rights Reserved. Page 21

27 Newington Smart Home ZBM Fault on 30 April 2011 Report Normal Operation (ZB Battery Voltage is quickly stripped to 0V on a daily basis) Faulty Operation (ZB Battery Voltage gradually falls to 0V on a daily basis) Figure 1: This graph shows the difference in battery voltage profiles before and after the system was briefly put into Manual mode. The red Bc1BatVtg curve, which shows ZBM battery voltage, falls quickly during operation on 17 February, but falls gradually during all times after this. This did not pose an issue for some time, as the battery was discharged for an adequate period of time each day to allow the battery s voltage to reach 0V naturally by slowly discharging. However, on 14 April, after discussions with Ausgrid, a new charge/discharge cycle was implemented for the battery, aiming to stop the system from requiring grid power towards the end of charging the electric vehicle. By this time, other settings of the operation of the whole ESS system had been optimized so that charging the electric vehicle was now the only reason for importing grid power. This new cycle included increasing the charge time to 9 hours, and reducing the discharge/strip time to approximately 12 hours. While the battery s state of charge (SOC) always reached 0% after about 3 or 4 hours, this did not necessarily mean that the voltage reached 0V (refer to Figure 2). In fact, charging the battery for 9 hours, produced about 50% SOC, and evidently, 12 hours was not long enough to fully empty the stacks of all zinc, meaning the voltage never reach 0V after 15 April. This went unnoticed until 30 April when the fault occurred. During this 15 day period, zinc built up in the stacks until an internal short occurred, increasing temperature in the ZBM, which expanded the gases within, thereby forcing the leak through the relief valves (refer to Figure 3), and also causing the small bromine gas leak that went unnoticed by the family RedFlow Limited. All Rights Reserved. Page 22

28 Newington Smart Home ZBM Fault on 30 April 2011 Report SOC reaches 50% every day, and ZB Battery Voltage never falls to 0V SOC reaches 35% every day, and ZB Battery Voltage gradually falls to 0V Figure 2: The ZBM charge/discharge cycle was changed on 14 April see the differences in the mauve Bc1Soc curve, which shows ZBM battery SOC, between 13 April and 14 April. ZBM voltage, shown by the red Bc1BatVtg, reached 0V last on 14 April, after which it only reached approximately 35V. Figure 3: Leaked electrolyte near the pumps. Figure 4: Electrolyte spray was released from the red pressure relief valve in this picture RedFlow Limited. All Rights Reserved. Page 23

29 Newington Smart Home ZBM Fault on 30 April 2011 Report Implemented Changes to Operation In response to this fault, RedFlow has implemented several changes to its operations to ensure a repeat does not occur. These include: Development of updated BC firmware that includes monitoring of internal resistance and a fault trip (similar to the currently-implemented leak fault trip) if the internal resistance falls below a specified value. This signals a short in the battery stack is soon to occur, but will discharge the battery before shutting down the system. This upgrade will be implemented by RedFlow s mechatronics department in the near future. Current and future ZBMs are now being upgraded to include a larger relief valve (refer to Figure 5) and a leak containment chamber. This will mean a larger capacity to release leaks of electrolyte and gas to stabilize the battery. These leaks will also be contained within the new chamber, thereby further reducing the risk of leaks outside the ZBM enclosure. The chamber will also utilize a carbonated gas filter to prevent the smell of bromine from reaching the surrounding area. There will be a leak detector in the chamber to trigger an immediate shut down of the system. Figure 6: An exploded view of the Generation 2 ZBM (The model currently installed in Newington). The relief valve on the Gas Handling Unit (GHU) is highlighted with the yellow arrow. Figure 7: An exploded view of the Generation 2 GHU on the bromine side of the battery. The relief valve is shown as Part 7. Inclusion of ZBM battery voltage to daily monitoring graphs of the Newington system. Comparisons of all BC parameters to correct values on a regular basis. A company-wide awareness campaign regarding the importance of regular stripping, battery voltage monitoring and BC parameter monitoring, for example, of the flags hex value RedFlow Limited. All Rights Reserved. Page 24