GSA's Cool Coup at the Philadelphia Custom House

August 28, 2007

As reported in the Summer 2005 edition of FEMP Focus, FEMP performed a study of prospective load management and demand response approaches for the General Services Administration's (GSA) Philadelphia Custom House in early 2005. GSA adopted the key recommendation and, through a few targeted operational changes and almost no capital cost, saved nearly $70,000 in demand payments in 2005-2006 and more than $100,000 (almost 15 percent of the facility's annual electricity bill) in 2006-2007.

GSA pays more than $28 per kilowatt (kW)—two to three times the national norm—in demand charges for the 570,000 square foot Custom House, and is also subject to a demand "ratchet" such that 80 percent of its summer peak power draw (i.e., its highest single 30-minute interval reading between June and September) becomes its minimum billed demand for each of the next eight months (October through May).

Since the Custom House generally experiences a summer peak of about 2,000 kW, this means that GSA is obligated to pay for at least 1,600 kW during these off-peak months. However, the facility is a conventional federal office building with a low load factor, and barely reaches peaks of 1,000 kW from December to March. At more than $28 per kW, the Custom House regularly pays its utility (PECO Energy) over $15,000 per month during those four months (as well as additional sums in the "shoulder" months of October, November, April, and May) for power it does not even draw.

With this in mind, GSA commissioned FEMP to conduct a study on the potential to cost-effectively reduce its peak demand. The central component of FEMP's recommendation was a "pre-cooling" strategy where GSA would turn on its chilled water plant very early in the morning (as opposed to the usual 6 A.M.) on hot summer days. In addition, FEMP recommended that the chilled water valves in the building's roughly one thousand perimeter induction units be tripped to a "fail-open" position during these early morning hours so that the facility would actually be somewhat over-cooled. The idea was to utilize the circa 1934 building's substantial mass as a thermal storage medium, which could then absorb heat and provide cool-temperature radiation throughout the day, mitigating the customary afternoon power peak.

GSA adopted this strategy, and working with their operations and maintenance contractor, Brooks Range Contract Services, they developed a multi-part plan to reduce the building's peak through early morning pre-cooling and afternoon "demand-limiting." The key elements are:

  • If the outside air exceeds 70°F at 2 A.M., one of the facility's two 650-ton chillers is turned on and programmed to produce 42°F chilled water;
  • All induction unit chilled water valves are set to a full-open position during the early morning;
  • At 9 A.M., the chilled water temperature is raised to 46°F and induction unit control reverts to the tenants (the units have no re-heat coils but the unit controls can be set towards "warmer" to reduce or eliminate the flow of chilled water through them);
  • If demand reaches 1,500 kW and is still rising by 12 noon, the chilled water temperature is raised again, to 48°F;
  • Only one of the two 650-ton chillers is allowed to operate at any given time.

In the beginning of summer 2005, the Brooks Range team executed the strategy manually, using control system overrides for chiller operation and bleeding the air out of the pneumatic lines to open the induction unit valves. Once the team gained confidence in this strategy, the building's controls contractor was called in to help automate it within the energy management control system (installed in 2003 as part of a Super Energy Savings Performance Contract).

As a result, the operations team was able to keep the facility's peak demand down to 1,766 kW over the summer (defined by the PECO tariff as June though September), as opposed to the 2,050 kW or higher that would likely have been reached. GSA benefited directly from the reduced demand in the summer, saving an estimated $26,000 (see Figure 1, below) in those four months alone.

Custom House Demand Reduction and Savings - 2005-2006.

GSA reaped even greater savings from the reduced ratchet charges during the winter months. The ratchet clause set the minimum demand charge for the October through May bills at 1,413 kW (80 percent of the 1,766 kW summer peak). While the previous four summers' average peak was 2,080 kW, FEMP conservatively estimated that 2,050 kW would have been 2005's peak draw (this is a conservative estimate because the summer of 2005 was an unusually hot one in the mid-Atlantic). Since 80 percent of 2,050 is 1,640, this figure was used to estimate the ratchet savings – i.e., to represent what the billed peak would have been without the pre-cooling. The 227 kW reduction (1640 – 1413) translated to more than $30,000 in savings for the five months of December through April; additional ratchet relief in October, November, and May made for a total (including the $26,000 in direct summer months' savings) of roughly $68,000.

FEMP's preliminary sense was that there would be a slight – perhaps 5 percent – energy (kilowatt-hour) penalty for implementing the pre-cooling, since some of the "coolth" generated by the chiller and HVAC system in the early mornings would escape from the building envelope without generating comfort cooling. Unquestionably, this effect took place. However, the building's summer electricity usage does not seem to have gone up. Next to the most comparable recent summer (2002), summer 2005's usage was only 0.5 percent higher, despite the fact that it had 4.3 percent more cooling degree days. Moreover, a regression plotting the four previous summers' kWh consumption against the number of cooling degree days in each revealed that summer 2005's actual consumption was 2 percent less than what the model predicted. The facility's summer 2006 usage fell a remarkable 7.5 percent below the regression's prediction.

Why was consumption lower than expected? One reason might be that, since only one chiller was run at a time, average chiller operation was at much higher load conditions than in previous summers – this raises efficiency. Another small effect might be from the cooling towers, which purge their heat into 70 degree night air more efficiently than in 95 degree sunny conditions.

Photo of the General Service Administration's Custom House
General Service Administration's Custom House, Philadelphia, Pennsylvania

Despite the impressive savings, GSA was adamant from the start of the pre-cooling experiment that any decrement in occupant comfort would not be acceptable. Both FEMP and GSA were concerned that the pre-cooling would generate cold complaints in the mornings as tenants arrived at work, and hot complaints during the hottest afternoons while GSA held to only one chiller as part of the "demand-limiting" strategy. However, this was not the case. "Thermal complaints went down," asserts John Kleaver, the GSA building manager. "The tenants have never been happier," agreed Brooks Range's Rich Ponticelli, head of the operations team that implemented the pre-cooling. Corroborating these accounts were GSA's complaint logs, which showed that hot complaints went down from 41 in 2004 to 26 in 2005 (despite the much hotter summer of '05). Interestingly, cold complaints went down slightly also, from 10 in 2004 to 6 in 2005.

In previous summers, the fact that the facility's chilled water system was shut off at 5 P.M. and did not resume operation until 6 A.M. the following morning meant that some tenants, particularly the early birds, found the building too warm and humid upon arriving for work on the hottest summer days. The pre-cooling strategy appears to have resolved this problem – and without over-compensating. The 2 A.M. start time seems to have been an excellent choice by the operations team.

In sum, the Custom House's pre-cooling thermal storage experiment has been an enormous success. The GSA avoided almost $70,000 in demand charges during the first year (2005-6), all for the cost of a visit from their controls contractor and two solenoid valves. GSA manager Tom McGarry concluded at a "lessons learned" meeting that the GSA should "declare victory" and commit to repeating the strategy in the summer of 2006. Indeed, GSA was able to suppress demand even further in the summer 2006, from an expected 2,100 kW to 1,684 kW. This 20 percent reduction generated savings of roughly $103,000, along with approximately $6,000 in kWh savings from the 7.5 percent energy consumption reduction. Needless to say, GSA is implementing the strategy once again in 2007 and plans to continue it in future years.

For more information, please contact Phil Coleman, Lawrence Berkeley National Laboratory, 610-604-0170.