TMCnet News
Implementing Automated Demand Response [Chemical Engineering Progress](Chemical Engineering Progress Via Acquire Media NewsEdge) SMART GRID Low-cost automation and communication systems will be critical to the success of demand-response programs. Electric utilities and the independent system operators (ISOs) that manage the electric grid offer various demand-response (DR) programs to incentivize electricity customers to change their end-use loads upon request. Demand response was originally designed to reduce peak loads during hot summer days or to mitigate problems during emergencies. As the electric system incorporates more intermittent renewable electric supply, DR programs that allow customers to provide more-flexible demand are becoming important tools for managing the dynamics of the electric grid. Grid operators traditionally balance supply and demand manually to control the power flows between the transmission systems and the energy consumers. The ISO continually monitors both supply and demand in real time, as well as forecasted demand for the short term (an hour ahead) and medium term (a day ahead). When demand is expected to exceed supply, the ISO telephones the utility with instructions to increase output by a specified amount starting at a certain time for a prescribed period. The power plant responds, often by starting a peaking gas-fired generator. The ISO might also contact commercial building landlords and ask them to increase temperature setpoints to reduce power consumption for air conditioning. While this relatively simple approach to automation works for a residential or commercial building, industrial sites, especially chemical process industries (CPI) plants, present some unique challenges. One challenge is the communication between the electric generator and the industrial user. This article discusses the development of a DR communication and automation system that can be interfaced with existing industrial process control systems. Automating industrial demand response Research began in 2002 to develop a low-cost DR communication and automation system that would transmit electricity price and grid reliability signals, and that could easily be connected to existing industrial control systems. The technology was called Open Automated Demand Response (OpenADR) - open to distinguish it from other, proprietary automated DR efforts. The system uses a client-server communication architecture with an open application-programming interface (API) and a web server that communicates with remote software clients (Figure 1). In OpenADR, the client side of DR automation is deployed via software clients embedded in control systems or deployed with gateways. The DR Automation Server (DRAS) publishes signals from the utility or the grid operator to communicate with the remote client, and the client uses pull technology to request event data from the DRAS every minute. This system operates continuously year-round. The OpenADR Version 1.0 specification was officially published in 2009 and has been used in California since 2006 to provide about 250 MW of commercial and industrial automated demand response (I, 2). At around the same period, the National Institute of Standards and Technology (NIST) initiated coordination in the U.S. to identify, develop, and implement smart grid interoperability standards. These efforts at formal standardization of DR and distributed energy resources signals produced a set of communication specifications that is now known as OpenADR 2.0 (2, 3). OpenADR 2.0 (Figure 2) uses different terminology for clients and servers than Version 1.0 - servers are known as virtual top nodes ( VTNs) and clients as virtual end nodes (VENs). OpenADR 2.0 offers additional services to support standardization of both wholesale and retail DR market signals in the U.S. The original concepts and semantics, however, are the same, which will ease the future transition of systems that now use Version 1.0. Figure 3 illustrates the flow of information over an OpenADR network. An electricity price or another DR event signal is described in a software data model and translated into an OpenADR signal. In the example in Figure 3, a signal representing a dynamic price event, such as $0.20/kWh for the 2 p.m. to 3 p.m. hour, is sent over the Internet via a server, and may be sent a day ahead or an hour ahead of the event. The industrial facility's control system receives the signal and executes a preprogrammed strategy. The system has secure two-way feedback; the server tracks the signal to the client and sends a return signal confirming that the information has been read correctly (e.g"Got it - $0.20/kWh, 2-3 p.m."). Figure 4 illustrates the terminology related to the timing of a DR event. Control systems are given information about the start time and end time of an event or a price. Pending signals of the event start time and the duration of the event allow loads that need advance notice, such as motors and drives, to prepare and begin changing their state. Slow loads, such as thermal mass in buildings, may require several hours of notice to pre-cool and prepare for a hot afternoon event. Large industrial loads may also need several hours of notice to prepare to enter the active state at the start of the event. Pilot programs and field experience The development and deployment of DR automation systems has been underway for many years. More than 1,300 buildings and industrial facilities in California have installed OpenADR to participate in automated DR programs, providing about 250 MW of DR (2). The next two sections present the results of two field tests involving commercial buildings. Contingency market pilot In 2009, California conducted the Participating Load Pilot (PLP) program as a first step toward allowing DR resources to participate in the California independent system operator's (CAISO's) DR markets. The objective of this program was to assess the technical and financial feasibility of using retail DR resources as participating load (PL). A PL resource could participate in the day-ahead, real-time, and/or contingency response markets. Three facilities - a retail store, a local government office building, and a bakery - participated in the contingency market portion of the PLP program (4). These sites had been on day-ahead automated critical-peak pricing, and were selected because their hourly load profile and demand response were among the most predictable in the automated DR program. Each facility was equipped with 4-sec near-real-time telemetry equipment to provide whole-building power data used to monitor and evaluate the performance of the DR strategy. For each facility, hourly demand and load-curtailment potential were forecast two days ahead of a DR event, and this information was submitted one day ahead to the CAISO as an available resource. CAISO optimized these DR resources against all other available generation resources. When CAISO requested DR resources, the utility's OpenADR communications infrastructure delivered DR signals to the facilities' energy management and control systems. which executed the appropriate preprogrammed DR strategies (e.g., adjusting building temperature setpoints up or down). Figure 5 displays the 5-min load data and hourly forecasts for a PLP event that took place from 2 p.m. to 6 p.m. The DR strategy for this facility (the office building) was programmed such that four DR load-reduction levels were mapped onto four 1°F incremental temperature adjustments. At the start of the event, a 2°F adjustment was initiated and the load reduction was sustained during the entire four-hour event. The DR resource needed to be available within 10 min of receiving an event signal, and this time requirement was easily met. Regulation response pilot Regulation is a segment of the ancillary services market (Table 1 ) that is required for continuous balance of generation and load, whereby the provider of the service is equipped with automated controls that allow the ISO to request upward or downward changes in generation or load. Regulation is used to track and balance system-wide generator output with system-wide load on a sub-minute-bysub-minute basis. This requires extremely rapid automated responses from resources (primarily generators) that have been characterized and certified as meeting certain requirements, such as speed and level of response. Tests conducted by Lawrence Berkeley National Laboratory (LBNL) with the CAISO and a local utility examined whether DR resources such as electricity consumers could meet the requirements to replace generators in the regulation market and whether OpenADR could meet the communication speed requirements (5). Each participating site (two universities and an agricultural pumping station) was equipped with an OpenADR client connected to the OpenADR server. Figure 6 shows the architecture of one participant's system. In a typical regulation arrangement, the generator is con- nected through its energy management system (EMS) to the CAISO's automatic generation control system. They communicate over a secure private network using inter-control center communications protocol (ICCP), and the generator adjusts its power output in response to setpoint instructions from the AGC system. The field tests used a similar architecture for loads. The AGC transmitted setpoint instructions through a private network using ICCP, but instead of sending these signals directly (as it does to generators), it sent them to a DRAS, which is the server that communicates standardized DR signals. The DRAS converted the AGC signals to OpenADR signals and pushed these signals to the software-based OpenADR clients located at each site. The OpenADR clients received these setpoint instruction signals and triggered preprogrammed DR strategies corresponding to the setpoint set by the CAISO. Revenue measurement took place in the utility's meter data management (MDM) system. Tests evaluated the characteristics of demand response (Figure 7). The top graph of Figure 7 plots load over the 2 a.m. to 4 a.m. test period, during which building ventilation was manually increased every half-hour starting at 2 a.m. The bottom graph zooms in to provide a better view of the ramping. The purple line represents a conservative estimate of the ramp rate - 38 kW in 21 seconds. The project also examined the latency of the round-trip control signals. Incorporating an OpenADR server and clients to facilitate automation and convert ICCP to OpenADR using the public Internet added, on average, less than a 2-sec delay in the transmission of these signals to the facilities. The latencies after the OpenADR clients received the signals ranged from 2 sec to over 2 minutes. DR attributes There are many unanswered questions about how quickly industrial facilities can respond to signals from a utility or grid operator, as well as how frequently the load can respond, and for how many hours. The speed or response time for a load to be reduced depends on the latency of the controls and the nature of the load. In commercial buildings, lighting usually responds faster to DR signals than heating, ventilation, and air conditioning (HVAC) systems, which have motors and compressors with inherent inertia that ramp up and down more slowly. For slower responses, cooling systems have the potential to precool loads and offer more flexible services when the mass of the building is considered. The demand-response capabilities of controls in industrial facilities depend on the type of facility. For example, certain types of resources, such as agricultural water pumping stations, data centers, and refrigerated warehouses, can participate in ancillary service markets for fast responses. The industrial loads with slower responses may be good candidates for day-ahead DR markets. Table 2 lists seven attributes that grid operators and researchers consider when estimating the flexibility of DR resources to meet emerging grid needs. The cost to automate DR is a key consideration. It is often more cost-effective to automate large facilities with centralized energy-management systems than small, locally controlled ones, because the cost to install and enable the DR automation buys a much larger, centrally controlled load. The lowest-cost DR scenario would involve control systems that are designed and built to a common standard and have native software capabilities for receiving and responding to grid signals. DR automation costs A demonstration project evaluated the deployment of OpenADR technology in the Seattle, WA, area to reduce winter morning electric peak demand in commercial buildings. Table 3 shows the costs required to install the DR automation hardware and software. The two Target stores had the lowest average costs because their control systems had native OpenADR software. California utilities have been paying incentives of about $200-$400/kW to industrial facilities so they can invest in the automation technology needed to facilitate participation in DR programs. Buildings equipped with such automation that incorporates native OpenADR software can configure automated DR strategies for under $50/kW (4). Future directions Resources have been shown to be technically capable of providing fast demand-response services. The Internet can enable fast DR-based nonspinning ancillary services, which is critical for low-cost automation. It is likely that as the technology and market opportunities for DR continue to evolve, the costs for the DR resources will be much lower than those of conventional supply resources. The lowest-cost vision for the fiiture would be one in which facility loads can create multiple value streams at different time scales. Facilities would be able to automate once and use many times. For example, a load might participate in both summer and winter DR programs, fast and slow DR programs, as well as various ancillary services programs. However, some questions remain. How can DR automation platforms and end-use loads enable a broad set of grid transactions? Can a facility participate in more than one DR program? Will there be "fatigue" if DR events become frequent? More work is needed to develop decision-making and analysis tools for facility owners and managers to support automated demand response. New work funded by the U.S. Dept, of Eneigy (DOE) is starting to explore this area using an agent-based decision-analysis platform for grid integration. LBNL is developing an automated measurement and verification tool to support the transactive platform. We often describe the costs to automate the load in terms of the $/kW, but more research is needed on the costs to install and maintain these telemetry platforms over time, and to evaluate the economic value of transactive systems. E2J LITERATURE CITED 1. Piette, M. A., et aL, "Open Automated Demand Response Communications Specification," CEC-500-2009-063 and LBNL-1779E, California Energy Commission, PIER Program, Folsom, CA, and Lawrence Berkeley National Laboratory, Berkeley, CA (2009). 2. Ghatikar, G., et aL, "Analysis of Open Automated Demand Response Deployments in California and Guidelines to Transition to IndusOy Standards," LBNL-6560E, Lawrence Berkeley National Laboratory, Berkeley, CA (2014). 3. Holmberg, D. G., et aL, "OpenADR Advances," ASHRAE Journal, 54 ( 11), pp. BS16-BS19 (2012). 4. Kiliccote, S., et aL, "Northwest Open Automated Demand Response Technology Demonstration Project," LBNL-2573E, Lawrence Berkeley National Laboratory, Berkeley, CA (2010). 5. Kiliccote, S., et aL, "Field Testing of Automated Demand Response for Integration of Renewable Resources in California's Ancillary Services Market for Regulation Products," LBNL-5556E, Lawrence Berkeley National Laboratory, Berkeley, CA(2012). ACKNOWLEDGMENTS This work was sponsored by numerous organizations. Support was provided by the California Energy Commission Public Interest Energy Research (PIER) Program under Work for Others Contract No. 500-03-026, and by the U.S. Dept, of Energy (DOE) under Contract No. DE-AC02-05CH11231. We are also thankful for the support of Pacific Gas and Electric Co., Seattle City Light, the Bonneville Power Administration, and the DOE's Building Technology Office. MaryAnn Piette Si la Kiliccote Girish Ghatikar Lawrence Berkeley National Laboratory MARY ANN PIETTE is head of the Building Technology and Urban Systems Dept, and the Director of the Demand Response Research Center (DRRC) at Lawrence Berkeley National Laboratory (Email: [email protected]), where she specializes in commissioning, energy information systems, benchmarking, and diagnostics, and she develops and evaluates low-energy and demand response (DR) technologies for buildings. She developed demandresponse technology and the Open Automated Demand Response standard (OpenADR), which is being deployed to deliver over 250 MW of DR in California and throughout the U.S., and served as the first chair of the OpenADR Alliance. She has authored over 100 papers on efficiency and demand response. Piette holds a BA physical science and an MS in mechanical engineering, both from the Univ. of California Berkeley, and a licentiate in building services engineering from Chalmers Univ. of Technology, Sweden. SILA KILICCOTE is Acting Leader of the Grid Integration Group at Lawrence Berkeley National Laboratory (Email: [email protected]), where she is part of the automated demand-response team that is developing an automated communication infrastructure, integrating it with building control systems, and working with stakeholders to standardize the information model. Her areas of interest include characterization of buildings and demand reduction, demand-responsive lighting systems, building systems integration, and feedback for demand-side management. Kiliccote holds a BS in electrical engineering from the Univ. of New Hampshire and an MS in building science from Carnegie Mellon Univ. GIRISH (RISK) GHATIKAR is a deputy lead for the Grid Integration Group at Lawrence Berkeley National Laboratory ([email protected]), where he oversees the work on U.S. and international demand-response (DR) technologies, OpenADR standards, and related technologies. His background and experience are in information technology, standards, technology transfer, and business innovation for energy efficiency, OR, smart grid, and their field applications. He serves on the Board of Directors and as vice chairman for the OpenADR Alliance; on steering and technical committees for the Organization for Advancement of Structured Information Standards; and as a member of user groups to accelerate the adoption of smart grid and clean energy innovation in the U.S. and international markets. Ghatikar holds an MS in telecommunication systems, computer technologies, and infrastructure planning and management. (c) 2014 American Institute of Chemical Engineers |