2016 ASHRAE Winter Conference

The as-installed energy efficiency of unitary systems is much less than that of central systems, and the efficiency gap widens as systems age due to maintainability issues.  Energy engineers and service technicians use indirect indicators of equipment performance and make adjustments according to manufacturer guidelines and standard field practice, which varies with technicians’ level of experience.  Growing numbers of unitary systems combined with shrinking budgets result in deferred maintenance, and long-term operation of equipment at degraded levels. Energy efficiency is a metric that must be measured to be optimized.  This paper reports on field testing of continuous sensing of actual operating energy efficiency to control unitary equipment operating parameters, provide remote fault detection diagnostics and support maintainability. The energy efficiency of most unitary HVAC systems is much less than chilled water systems and limited cost-effective choices exist for increasing their energy efficiency.

In the U.S., there are 14.4 M electric-heated dwellings using 0.16 quad/year for heating in cold regions, e.g. ASHRAE climate regions IV and V. Conventional air-source heat pumps (ASHP) do not work well in such cold zones, due to high compressor discharge temperatures, large pressure ratios and inadequate heating capacities at low ambient temperatures. Consequently, noticeable portion of auxiliary strip heating has to be used to meet the building heating load. Two options of tandem compressors were studied, with one using two identical, single-speed compressors, and the other using two identical, vapor-injection compressors. The system modeling and laboratory testing results will be presented in the paper.

A 250 m², two story, residential home of the style typical of the Gaithersburg, Maryland area was constructed in 2012 to illustrate technologies employed to produce a netzero energy home. It functions as a laboratory to support the development and adoption of cost-effective NZE designs, technologies, construction methods, and building codes. The primary design goal was to meet the comfort and functional needs of the simulated occupants. The first annual test period began on July 1, 2013 and ended June 30, 2014. During this period the heating and air conditioning of the home was performed by a novel air-source heat pump that utilized a reheat heat exchanger to allow hot compressor discharge gas to reheat the supply air during a dedicated dehumidification mode.

The application of vapor injection to the compression process leads to a decrease in discharge temperature, extending the operating envelope to lower suction pressures. Additionally, it increases the coefficient of performance as well as the heating capacity under these conditions. Vapor injected compressors are therefore ideally suited for cold climate heat pump applications. This paper introduces a PI-type mapping of a dual port vapor injected compressor’s performance data trained with data from both, in-system testing as well as test-stand testing.  The in-system testing was conducted in a prototype cold climate heat pump, where injection mass flowrates and suction superheat were a result of the operating conditions.  In contrast, suction superheat of the test-stand data was fixed while injection flowrates were dictated by the test plan. These differences result in limitations of the mappings if the model is trained with only one of these sets.

Vapor injected compressors are one of the promising technologies for cold climate heat pumps because it enables compressor operation with a large range of evaporating temperatures at acceptable efficiency. However, the large operating range also makes it difficult to generate empirical maps that are applicable to most of its operating conditions. This paper discusses a method to create an empirical map using the Buckingham-PI theorem and estimates the resulting extrapolation uncertainty. Data from the laboratory testing of a cold climate heat pump were used to generate the map. The uncertainties from model random error and training data were calculated to examine if the model is reliable at extrapolation.