Demand Response

13 November 2008

Security of power supplies is a concern for the future. In the US power producers and business users are regulating peak energy use to match supply to reduce demand and the risk of power cuts. Wayne Wiebe explains how and some of the lessons learned

THE CALIFORNIA OF 2001 WAS FILLED WITH ELECTRICAL brown-outs, utility bankruptcies, and Enron shenanigans. In response, California’s energy governing body was looking for ways to reduce usage and make energy more reliable. It financed a study to ‘understand the technical opportunities of automating demand response and to remove technical and market impediments to large-scale implementation of automated demand response’ through Laurence Berkeley National Labs (LBNL). This Automated Demand Response (AutoDR) study mimicked dynamic electric prices and sought commercial and public customers’ reactions to rapid price fluctuations. LBNL asked Echelon to be one of the first participants in this study.

Echelon was chosen because its HQ is representative of highly integrated building automation systems, and for its development of the LonWorks® control networking platform widely adopted by the building automation industry. The campus’ LonWorks building automation system can lower electrical consumption within 10 minutes. The path to figure out what to shed and how to shed without adversely affecting our employees took time.

Demand response (DR) refers to mechanisms to manage the demand from customers in response to supply conditions, for example, having electricity customers reduce their consumption at critical times or in response to market prices. AutoDR uses automation to eliminate the need for human intervention to cause an action or set of actions to take place.

In its simplest form, DR could entail a phone call from a utility to a FM followed by personnel physically adjusting various environmental parameters. An AutoDR program could entail a notification via the Internet that is registered by the building automation system, which in turn initiates a series of actions and parameter changes that reduce energy consumption. In most instances, the systems most likely impacted by a DR program are lighting and heating, venting, and air-conditioning (HVAC). Given the US’ reputation as an energy waster, it is surprising that the US is furthest along in deploying DR programs. The US federal government estimates that in 2007 potential DR capability equaled about 3 per cent of total US peak demand or 21,900 MW. States with programs in place include all of the Northeast, Texas, and California.

There are clear differences between the implementation and infrastructure of the US and Europe. For instance Europe consumes 6,000 kWh per year, per capita and the US averages 18,000 of kWh per year. However, the EU peak demand is growing faster, at 2 per cent annually, than that in the US.

Most countries in the EU have just one electric utility, often state-owned; whereas there are some 3,000 separate load-serving entities operating in the US. In Europe, where DR is just now beginning for most countries, an essential technology of automatic demand response is commonplace – modern European buildings typically feature integrated building automation systems that can easily be leveraged by DR programs. It would be no surprise then to see the EU accelerating well past the US in DR capability.

Furthermore, the EU is also ahead of the US indeploying smart metering systems, which include time of use (TOU) capability, and can accelerate DR programs. One report estimates that very aggressive DR programs will avoid €50bn in new generation capacity and new transmission and distribution. This is equivalent to 150 medium sized gas fired power plants over the next 12 years.

States in the US with deregulated utility markets have turned over the marketing and deployment of DR programs for commercial and industrial consumers to aggregator companies. Aggregators sign agreements with the wholesale power provider organizations in these deregulated markets. This provider is the Independent Service Operator (ISO) in California. In California, the aggregator will commit to provide 50 MW of reduced peak demand during an event, and the ISO will pay for that power which was not used. In the US, aggregators may be national or regional in scope.

The LBNL study in which Echelon participated has been successful. In 2007, 152 Californian commercial and industrial customers participated in AutoDR and reduced their peak load by 25MW, representing about 34 per cent of their aggregate peak demand. Just one year before, 13 participating customers had reduced their peak load by 1MW, an average of about 11 percent; then a set of orders issued by the State's Public Utilities Commission in late 2006 led to the broad expansion of the program.

EnerNOC provides an example of a national aggregator’s system. When the ISO sees the need for peak reduction it will call an aggregator service center – in the case of EnerNOC, the Network Operations Center (NOC), and declare an event.

To initiate reductions at sites within its network, the NOC then sends a signal to each customer’s EnerNOC Site Server, an on-site installation that typically contains control relays and an iLON, a LonWorks device that monitors energy usage which EnerNOC deploys as a virtual TOU meter in its typical customer application.

“The EnerNOC Site Server and the iLON gives our NOC visibility into how energy is being consumed at each site within our network. With 5-minute interval data streaming to us in near real-time, we can actively manage our demand response assets to ensure the required load reduction is met,” said Jim Hutton Johnson, Marketing Manager at EnerNOC. Lights are turned off, HVAC setpoints are raised, and energy is saved. The NOC records how much energy is saved. At some interval, perhaps quarterly, the customer is paid for his reductions.

When Echelon was first asked to reduce power consumption as a part of the LBNL study, the notification for a DR event was 24 hours. Over the course of three years the notification period has been reduced to two hours in our present DR program. Furthermore, tests have indicated the buildings can shed 30 per cent of consumption in ten minutes.

The first attempt at reducing consumption followed a systems approach. We simply shut off all building systems which consumed large amounts of electricity. The plan for the first couple of events was to shut off all the lights that were near windows, and one or two of the three roof top air conditioning units. Although not very elegant, this worked well by one measurement: consumption was reduced by 60 per cent during our second event. However, systemic change worked poorly in the eyes of uncomfortable employees:

Clearly a different approach was required to reduce consumption. We started looking for functional changes. The device-level control allowed by the highly integrated and flat LonWorks system allowed us to create an energy conservation mode. The conservation mode included raising the building temperature set point, reducing the duct static pressure. After programming and verifying the function of HVAC systems in this mode we began addressing problem spots in the building.

The device level control at Echelon allows addressable VAV’s which are tied into the motion sensors used to turn lights on and off when rooms are occupied – essentially giving us room by room DR control. The fine level of control allowed trouble shooting of hot spots at the VAV level.

Employees, now much happier, provided crucial feedback. People will remain engaged, and are willing to participate provided they are informed, and their communications are acknowledged as valuable. When the weather grows hot we send out notes that point out the possibility of a demand response event. After an event we calculate our savings and let people know how much we are reducing greenhouse gases and how much we have saved the company. We have always had a comment box at Echelon. Complaints are addressed promptly if only to say they cannot be corrected to ensure a feeling of ‘engagement’ between facility management and employees.

Six years of DR experience have caused us to look more closely at other systems that have a peripheral effect. We have implemented a ramp-out at the end of an event - a gradual return to the normal operating parameters instead of allowing the systems to work twice as hard to return to normal operating parameters at the end of the day when unoccupied mode is only an hour away. The effect of the ramp-out is purely financial – it saves Echelon money. We have also lowered the hot water loop temperature and closed the
automated blinds in the conference rooms.

Echelon’s DR participation has lead to reductions in annual electric cost of mid-single digit percentages over each of the past three years. This is significant if put into perspective. Eight years ago, when the first new building was undergoing continuous commissioning, costs were brought down annually in the mid-teen percentage year over year. The DR program forced us to work hard to find further system efficiencies. Fortunately the LonWorks control system allowed many adjustments and provided concise feedback. Our experience has taught us to:

....pay attention to customer/employee feedback and be flexible in the response.

....without buy-in, even the best strategies are doomed to failure.

....the level of control provided by being able to address individual devices ensures the absolute minimum disruption to the people and maximum opportunity for reducing consumption.

Our lessons were learned in a country and building automation environment that in many ways is significantly behind the EU. The EU has higher energy costs, more energy deregulation, a rapidly growing base of smart metering, and more building automation. That means that Demand Response that began in
the US with great success is likely to become the next great way to lower cost and save energy across the EU.

● Wayne Wiebe is Vice President, Kenmark and is currently responsible for the on-site management and operation of Echelon's corporate campus in San Jose.