|
|
|
|
Using AMI Data to Improve Direct Load Control Programs
By Christopher Dyson
|
|
|
Utility Direct Load Control (DLC) programs, which have become increasingly popular in recent years, face a number of program planning and evaluation challenges that could be mitigated with access to more detailed customer metering data from Meter Data Management Agent (MDMA) systems. This article discusses the different ways this MDMA data could be used to improve DLC programs, including more efficient program planning and recruitment, more accurate estimation and verification of load reduction impacts, and less expensive system maintenance. Some of these benefits are being realized now by utilities while others are still in the concept stage.
DLC Programs in a Nutshell
DLC programs are a key component of the demand response portfolios of many utilities and Independent System Operators (ISOs). They provide valuable peak load relief by reducing customer electricity usage during times of peak system demand. The most common way of doing this is to send a paging or radio signal to either a switch mounted on the compressor of an air conditioner or to a “smart thermostat.” The switch will force the air conditioner to cycle on less frequently than normal or, in the case of some programs, totally turn off the air conditioner for a number of hours. Smart thermostats will typically reduce air conditioning usage by raising the thermostat setpoints, although some smart thermostats can also cycle an air conditioner like a switch.
While most DLC programs only cycle air conditioners, some programs also cycle electric water heaters and pool pumps. Many DLC programs limit participation to only residential customers while others also recruit small commercial customers. Some programs allow customers to easily “override” a control event by pushing a button on their thermostat, while other programs intentionally make override difficult or impossible. To get customers to participate, most programs pay customers annual financial incentives ranging from $20 to nearly $200 per year, depending on the frequency of the cycling they are willing to endure or the size of their air conditioner.
Making DLC Program Planning and Recruitment More Efficient
When it comes to recruiting customers for their DLC programs, most utilities rely on very blunt and often unsatisfactory billing data analysis techniques. For example, most utilities will take monthly residential billing data and then use some kilowatt hour (kWh) threshold as a rule-of-thumb for estimating whether a customer has a central air conditioner or not.
These techniques do an adequate job of finding which customers have central air conditioners. However, what is most important for DLC program managers to know is which customers are using their air conditioners during the key peak days when system capacity is tightest. And more sophisticated DLC program managers want to know how many of these key customers are connected to overstrained feeders or substations where DLC programs could provide valuable targeted distribution system relief.
Having access to detailed customer metering data from an MDMA system would allow utility DLC programs to find and target these key customers. At the same time, the program would save money by not paying financial incentives to “free riders” -- customers who do not use their air conditioners frequently or during the most important periods of peak system usage. By saving money in this way, DLC programs could offer larger financial incentives to recruit these key customers.
Making DLC Program Load Reduction Impact Estimates More Accurate
Knowing how much peak load reduction their programs are achieving is crucial for DLC program managers. In the past, some DLC program could get away with “back of the envelope” calculations – such as 1 kilowatt (kW) of peak load reduction per cycled residential air conditioner – to estimate program load reduction effects. Yet increasingly, ISOs and utility system monitors and planners are insisting on more accurate estimates of these effects. Some ISOs pay DLC programs to serve as spinning reserves or balancing energy resources and they want to be sure that they are getting what they pay for. On summer afternoons when many utility grids are at the breaking points, utility system monitors want to be fairly confident that when they initiate a control event for a 50 megawatt (MW) DLC program, they will be getting 50 MW of peak load relief. Finally, there is growing evidence that these back-of-the-envelope estimates of peak load relief can be very inaccurate.
Yet developing more accurate estimates of DLC program peak load reductions can be very difficult and expensive. To estimate peak load reductions for such programs, evaluators usually have to install data loggers on the air conditioning equipment of a sample of participating customers and then collect these loggers at the end of the cooling seasons. These data are then run through analytical models, such as “kW load” and “duty-cycle” models, to estimate the load reduction impacts of the program. There are significant costs involved in this process including recruiting customers for the measurement and verification (M&V) sample, compensating them for the inconvenience, and then sending out technicians to install and then later retrieve the data loggers.
Access to detailed customer interval meter data from MDMA systems would benefit load reduction analysis in a number of ways. First, it would greatly reduce the cost of analysis since data would be accessed through the MDMA system rather than through expensive field collection processes. Second, it would allow for more flexible sampling designs. The problem with M&V customer samples based on on-site data loggers is that they are static. Most utilities expand their DLC programs in stages over time, and as a program moves into new areas of the utility’s service territory, weather conditions or demographic characteristics may make the old M&V sample unrepresentative of the new program participants. Having access to MDMA data would easily allow sampling of the new DLC program areas. These data would also allow for more customized sampling strategies such as for certain customer types of interest or customers located in the area of certain vulnerable feeders or substations.
However, in order to effectively replace in-field data loggers, MDMA systems would have to be able to provide data of the same quality as these loggers. Currently, relatively-inexpensive data loggers can store up to a year of one-minute amp readings for each customers. Certain analytical methods for estimating DLC program load reductions – such as “duty cycle” models – are most effective when using data over these long time periods and at these precise time intervals. Another potential concern is that while data loggers provide amp readings for only the air conditioners, most MDMA systems would likely provide less precise household-level data. However, recent analyses by KEMA indicate that household level metering data can be reliable as inputs to these analytical models.
Xcel Energy Case Study: Reducing DLC Program System Maintenance Costs
While the previously discussed benefits of MDMA systems for DLC programs are only at the concept stage, at least one utility -- Xcel Energy -- has been using its Automated Meter Reading (AMR) system to produce tangible benefits for its DLC program. Xcel has leveraged the data from its AMR system to make its system maintenance process much more efficient and cost effective.
Air conditioner control switches break down over time and if older DLC programs do not do regular maintenance, they are in danger of losing much of their peak load reduction capacity. However, DLC program managers face the additional challenge of not knowing which of their switches are not working. Due to cost constraints, many DLC programs have installed control devices that lack two-way communication capabilities. Therefore, there is no easy way for program managers to know how many of their devices are receiving the activation signals. Not surprisingly, switch failure rates increase as the switches age. Yet many legacy DLC programs do not have reliable records of the age of their devices. And even when they do have reliable records, as Figure 1 shows, failure rates for switches as old as 15 years are still below 40 percent. This means that, wholesale replacement of switches of a certain age may not make economic sense.
Figure 1
Failure Rates for Air Conditioner Control Switches as They Age
|
|
|
|
|
|
When Xcel Energy first initiated maintenance procedures for their Minnesota residential DLC program, Saver’s Switch, in 1998, their maintenance strategy was based on the age of the switches. The utility would send out technicians to the older switches in order to inspect wiring, to inspect the external and internal condition of the switch, and to run control test simulations to verify the functionality of the internal components.
Yet this age-based maintenance strategy proved to be very costly. Since only a minority of the older switches would be failing, the Xcel Energy technicians would have to make multiple visits to the same switch until it failed. Although this age-based maintenance strategy did discover 4,000 bad switches and recovered 4.5 MW of lost load reduction capacity, the field work took 9 months and the cost per recovered kW was $599.
Between 1997 and 2001, Xcel Energy implemented a CellNet fixed wireless AMR network with 1.4 million electric meters installed in parts of Minnesota and the Dakotas. The Saver’s Switch program staff decided to see whether or not this new AMR system could help reduce the program maintenance costs. In 2001 they developed an AMR-based maintenance strategy. They selected a test group of program participants, pre-arranged a modified AMR schedule to capture the electric usage data from the customers in this test group, scheduled a test for a day with weather similar to a typical control event, and then initiated the test control event. After this, they analyzed the meter data from the AMR system to see which switches had reacted properly to the control event. Finally, they forwarded the list of failed switches to their field technicians for site inspection and switch replacement.
This new AMR-based maintenance strategy proved to be much more cost-effective than the age-based maintenance strategy. As Table 1 shows, the Summer Saver’s program was able to test many more switches and annual time in the field was less than half of what it once was. The overall cost per kW of recovered load reduction capacity was about one quarter of what it once was.
Table 1
Comparative Maintenance Metrics
Age-Based vs. AMR-Based Strategies
|
|
|
|
|
|
Other utilities that have both AMR systems and DLC programs are now looking to adopt similar AMR-based maintenance strategies. For example, Pacific Gas and Electric (PG&E) is in the midst of large expansions of both its AMR systems and its DLC program. By the end of 2011, it expect to deploy about 10.3 million new “smart” meters and by the end of 2010 it also plans to have about 400,000 customers in its SmartAC DLC program. For instance, PG&E is looking to adopt an AMR-based maintenance strategy for the SmartAC program that is similar to the one that Xcel Energy had adopted.
Contact the author at cdyson@kema.com.
|
|
|
Download the April 2008 Issue
Use the link below to download the PDF of the full issue of the April 2008 Automation Insight for the complete print versions of the articles.
|
|
|
|
[download] Automation Insight - April 2008 (.pdf 157 kb)
|
|
|
|
|
|
|
|
|
|
