November 15, 2014


This report covers the long-term performance of mini-split heat pumps (MSHPs) in Massachusetts (Zone 5A); it is the culmination of up to three years’ worth of monitoring in a set of eight houses. This research examined electrical use of MSHPs, distributions of interior temperatures and humidity when using simplified (two-point, one per floor) heating systems in high-performance housing, and the impact of door open/closed status on temperature distributions. The use of simplified space conditioning distribution (MSHPs) provides significant first cost savings, which are used to offset the increased investment in the building enclosure.

Executive Summary

This report covers the long-term performance of mini-split heat pumps (MSHPs) in Massachusetts (zone 5A); it is the culmination of up to 3 years’ worth of monitoring in a set of eight houses. This research examined electricity use of MSHPs, distributions of interior temperatures and humidity when using simplified (two-point) heating systems in high performance housing, and the impact of door open/closed status on temperature distributions. The use of simplified space conditioning distribution (MSHPs) provides significant first cost savings, which are used to offset the increased investment in the building enclosure.

A literature search was conducted on two topics: MSHP performance and single-point/simplified heating distribution.

Overall, this project demonstrated that simplified space conditioning distribution using MSHPs can provide excellent performance, as shown in many houses. However, there are some cases and situations that designers should be aware of as potential failures. Occupant operation can have significant impacts on performance. Key results from the monitoring include the following.

  • General MSHP operation patterns. Some patterns were seen over many of the monitored houses; it was useful to confirm expected behavior for the equipment. When a constant interior set point is used, the MSHP modulates up and down with outdoor temperature, running almost continuously throughout the winter to meet load. The first- floor unit provides the majority of the heating (compared to the second-floor unit), due to thermal buoyancy. Conversely, in summertime, the second-floor unit often provides the majority of the cooling. As would be expected in a zone 5A climate, heating consumption far outweighs cooling consumption.
  • Equipment capacity and sizing. There were no cases where there were issues with equipment sizing or lack of capacity, indicating that these cold-temperature heat pumps are a viable strategy as a single heat source in cold climates. This was confirmed by the monitoring: the MSHPs seldom hit maximum power draw, indicating substantial excess capacity even during worst-case winter conditions (much colder than local design temperatures). This is consistent with the installed capacity of the equipment: the oversizing (compared to calculated loads) ranged from 150% to 200% in most cases. Oversizing of MSHPs can actually be beneficial: they modulate their capacity, and their highest efficiency is obtained when the unit is running at the lower end of its capacity range.
  • Normalized use versus simulations. MSHP heating use was tabulated for the various houses, normalized by heating degree days (HDDs), and compared with simulation predictions for heating use. There is considerable scatter in the results: correlations varied from 57% above to 26% below the simulation prediction. Possible explanations were provided for various individual houses being above or below simulation predictions. Given the limited correlation between actual and simulated use, it is difficult to draw any conclusions on the accuracy of the energy model.
  • Normalized use (kWh/ft2∙HDD at 65°F [HDD 65]). The heating electricity use was also normalized by HDDs and square footage. Average consumption was 0.00030 kWh/ft2∙HDD 65, which is reasonably close to simulations predictions (0.00028 kWh/ ft2∙HDD 65). This information also provides a comparison metric for other houses heated with MSHPs.
  • Interior temperature distributions (4°F difference). The Air Conditioning Contractors of America recommends a maximum 4°F difference within a home or zone (highest minus lowest temperature); the temperature data were evaluated using this criterion. Results spanned a wide range: looking specifically at wintertime operation, results ranged from 96% of hours within the 4°F band, to only 19% of hours. However, some weaknesses of this metric were pointed out. In addition, summertime data were analyzed; they indicate that summer conditions are less challenging than winter conditions for simplified distribution, at least given the solar gains (glazing ratios and solar heat gain coefficients) in these houses.
  • Door operation effects on bedroom temperatures. Previous work showed that bedrooms during closed-door hours had greater differences between hallway and bedroom, compared to open-door hours. The current data were analyzed; however, limited conclusions could be drawn. Many houses had very few closed door hours, which might reflect actual operation, or instrumentation issues. Another house operated their MSHPs in an on-off manner (instead of constant set point), resulting in few usable data for evaluating door operation.
  • Thermal buoyancy effects (use of single MSHP on first floor). Two small houses were equipped with a single MSHP on the first floor, due to their small loads. However, during the first summer, the second floor did not cool down to set point (10°F or warmer than the first floor), even with the use of transfer fans. These issues are clearly due to thermal buoyancy: conditioned air rises from the first-floor unit in the winter, but it stays on the first floor during the cooling operation. An additional MSHP was retrofitted to the second floor, correcting this issue.
  • Open-plan first-floor temperature distributions. In general, open-plan first floors had few issues. The few exceptions were due to geometry and thermal buoyancy (an open stairwell intercepting heating air before it could reach across the space), and localized air leakage (dryer vent), resulting in a single cold room.
  • Bonus room comfort issues at Easthampton: One house experienced comfort issues that are instructive: the owners complained of a bedroom suite and a bonus room that were consistently cold in wintertime. A constant set point was used, but leaving doors open was not compatible with their lifestyle and schedule. Monitoring confirmed that extended winter periods with closed doors resulted in temperatures in the high 50s in the bedroom suite, and high 40s in the bonus room. This house is larger than other monitored houses (2300 ft2, versus 1100–1700 ft2 for others); in addition, it has unfavorable geometries in the problem areas. The bonus room had severe conditions, of exterior temperatures on five of its six sides. Calculations indicated that this was not an equipment undersizing issue, but a heat distribution issue. The problem was resolved by installing a 3:1 (indoor units: outdoor unit) MSHP, with indoor heads in all three bedrooms.
  • Temperature setbacks (on/off operation): Previous work has shown that deep temperature setbacks of simplified heating systems can exacerbate temperature unevenness issues. One homeowner complained of temperature unevenness; when the data were examined, it was clear that they operated their MSHP in an “on-off” manner, rather than using a fixed set point. This resulted in wide swings in interior temperature (between 60°F and 70°F+). The electricity use showed many hours with the MSHP running at maximum capacity, followed by periods with the unit shut off. When operated in this manner, the MSHP is heating at its least efficient (maximum output) state. Electricity consumption was a high consumption outlier; when compared with simulations, it was the worst-performing house (heating use 57% higher than simulation).

1 Introduction

This report covers research on the long-term performance of mini-split heat pumps (MSHPs) in a Northeast climate (U.S. Department of Energy [DOE] zone 5A); it is the culmination of up to 3 years’ worth of monitoring in a set of eight houses. This research examined electricity use of MSHPs, distributions of interior temperatures when using simplified (two-point) heating systems in high performance housing, and the impact of door open/closed status on temperature distributions. In addition, the builder’s real-world experience with these systems (including homeowner comfort issues) is discussed.

1.1 Problem Background

Conventional furnaces and split-system air conditioners are grossly oversized for many current high performance houses. It is common for such houses to have design loads of 12–18 kBtu/h; in comparison, 40 kBtu/h (nominal) is the smallest common furnace size. Conventional split system cooling systems start at 18 kBtu/h (1.5 tons), and high efficiency systems are often unavailable below 24 kBtu/h (2 tons). Holladay (2011) discusses the problem of selecting space conditioning equipment for low-load houses, and discusses various solutions.

Reduced mechanical system cost is often given as one of the benefits of increased building insulation and airtightness. Unfortunately, the first cost savings from reducing capacity with a conventional split system or furnace by 1 ton (to the smallest available) are modest, as most of the cost is in the labor of installation.

Inverter-driven ductless heat pumps (DHPs) or MSHPs offer a promising answer to these issues. They have been widely installed in Asia and Europe for more than 40 years, and have rapidly gained traction in North America. They are commonly available in sizes from 9 kBtu/h to 18 kBtu/h (0.75–1.5 tons), with some larger sizes as well. The equipment is more expensive on a per-ton basis, compared to fully ducted conventional systems. However, they offer significant installed cost savings relative to conventional system, when distribution costs are accounted for. Many MSHPs have a rated coefficient of performance (COP) typically at the top tier of commercially available equipment; they also offer variable-speed compressors, which render them even more efficient at off-peak conditions, and reduce the downsides of oversizing equipment. More recently, manufacturers have offered MSHPs that maintain their nominal heating output at 5°F or below, making them viable as a sole source of heating in cold climates, without a backup heating system.

The remaining—and substantial—challenge for wider deployment of MSHPs is the uncertainty surrounding thermal comfort in houses without distribution of hot and cold air to every room. Installing MSHP heads in each bedroom will increase costs sufficiently to negate their price advantage over conventional ducted systems.

1.2 Builder and Research Project Background

Transformations, Inc. is a residential development and building company with a proven track record of delivering high performance superinsulated housing at a cost-effective price point in a variety of Massachusetts markets (DOE zone 5A). Its production houses commonly include renewable energy systems, and have often achieved net zero and net positive performance. Building Science Corporation (BSC) has been working with Transformations since 2009 under the Building America program, on a variety of single-family and duplex projects (see Ueno et al. 2013a). Transformations, Inc. was named DOE Challenge Home 2013 Winner (Housing Innovation Awards), in both the custom home and production home categories.

Part of Transformations Inc.’s strategy of producing high performance homes without a significant cost increase is to offset the cost of upgrading the building enclosure/shell by reducing the size and cost of the mechanical systems. The builder uses MSHPs in its production work, typically with a simplified distribution system, of one indoor head/unit per floor. This has proven to be a very successful strategy in many of its past developments; however, many practitioners feel that additional research is warranted on the distribution of heating and cooling from point or simplified sources, and its effect on occupant comfort.

Therefore, monitoring equipment was installed at two Transformations, Inc. communities. Four houses at the Devens Green Zero Energy Community (Harvard, Massachusetts) and four at The Homes at Easthampton Meadow Zero Energy Attainable Community (Easthampton, Massachusetts) were selected for instrumentation. Further information on these communities and houses can be found in Appendix B and Appendix C. The instrumentation package included temperature and relative humidity (T/RH) measurements in several interior locations, electricity use of the MSHPs, door open/closed status, and exterior T/RH. The door status was recorded because it appears to have a strong effect on the temperature distributions of single point or simplified space conditioning systems. Details on instrumentation can be found in Appendix A.

1.3 Relevance to Building America’s Goals

Given the Building America goals of reducing home energy use by 30%–50% (compared to 2009 energy codes for new homes and pre-retrofit energy use for existing homes), Transformations Inc.’s houses are a demonstration that this type of performance is achievable cost effectively on a production basis. Providing research, validation, and guidance on the use of MSHPs and simplified mechanical systems is useful to support this builder in continuing construction.

This research on MSHPs and temperature distributions addresses a Building America Critical Path Milestone, as described in the document “Building America Critical Path Innovations Leading to 50% Savings” (NREL 2013). This research falls under the category of Space Conditioning, under Distribution System Solutions with Negligible Heat Losses: “Document distribution of T/RH distribution among rooms, utilizing MSHPs or equivalent as primary system, assuming economics prevent installation of a unit in every room.”

1.4 Tradeoffs and Other Benefits

The most obvious benefit to research on simplified heating systems such as MSHPs is the cost implications of reduced scope/size mechanical systems. As discussed previously, this strategy works in tandem with the improvements in enclosure performance. This is discussed in more detail in Section 3.

An additional advantage of MSHPs is that temperature zoning can be achieved, based on the number of heads available. Temperature variations between the first and second floors due to thermal stratification are a common problem.

Another cost reduction comes from the elimination of natural gas service to the house. As energy demand for space heating drops in high performance homes, the cost of installing and maintaining gas distribution becomes harder to justify. When monthly gas service charges and increased mortgage cost are counted as part of the heating cost, heat pumps and additional photovoltaic power can be more cost effective than a gas furnace. This is true even in cold climates (e.g., Massachusetts), and even when the furnace would use somewhat less energy on a source (primary) basis.

Finally, the elimination of burning fossil fuel within the house essentially removes safety risks from combustion byproducts compromising indoor air quality. Additional savings are achieved by eliminating the chimney or combustion venting, as well as any need to supply combustion air.

1.5 Research Topics and Research Questions

Key research questions include the following:

  • What range of temperatures is experienced in bedrooms of homes heated by point sources? As subsets of this work, what are the effects of door open/closed status, floor-to- floor thermal stratification, and house geometry on these temperatures and occupant comfort?
  • What is the typical heating balance point for a selection of these superinsulated houses? In this case, the term balance point means the outdoor temperatures above which no heating is required.
  • What are the electrical power consumption characteristics of MSHPs used in a cold climate, including monthly aggregate consumption, and consumption as a function of temperature?

2 Background and Literaure Search

The background and literature search section is divided into two interrelated subjects. First, an overview is presented on DHPs or MSHPs, from the current literature. Second, the topic of single-point or simplified heating distribution is discussed; there is some overlap between these two subjects, but this division provides some structure.

2.1 Ductless Heat Pumps/Mini-Split Heat Pump Background

2.1.1 Winkler: Laboratory Testing of Mini-Split Heat Pumps

Winkler (2011) described laboratory testing of two MSHP units (Fujitsu 12RLS and Mitsubishi FE12NA; both 12 kBtu/h nominal capacity). The team developed detailed performance maps, expanding on available performance data from the manufacturers. Heating and cooling testing was done under steady-state and cycling conditions.

On an installed per-ton basis, MSHPs are more expensive than conventional systems. However, they have very high rated efficiencies (25 seasonal energy efficiency ratio [SEER] or higher, compared to 18 SEER and higher for conventional systems). One reason for this testing was to validate those ratings under a wider range of conditions than the standardized tests.

Test variables included outdoor temperature, interior unit fan speed, and exterior unit compressor speed. The team found that experimental data matched manufacturers’ reported values. Maximum capacity measurements indicated that the units had even greater heating capacity than stated in manufacturers’ data, in particular, at cold ambient temperatures.

Under low and intermediate loads (part load conditions), both MSHPs had higher efficiencies than the conventional comparison high SEER forced-air heat pump systems. However, at peak load, the conventional systems had slightly higher efficiencies (by 10%–25%).

The author noted that MSHPs have a wide range of compressor speeds; therefore, they often do not cycle on and off, instead modulating their capacity to meet load while running constantly. This stands in contrast to conventional systems: even with two stages of heating/cooling, they will typically be sized to cycle on and off. The low, continuous operation of MSHPs is the most advantageous and efficient operating mode.

2.1.2 Baylon: Northwest Ductless Heat Pump Pilot Project

Baylon et al. (2012) reported on the results of the Northwest Ductless Heat Pump Pilot Project, under the auspices of the Northwest Energy Efficiency Alliance. This project involved the field monitoring of 95 homes retrofitted with MSHPs throughout the Pacific Northwest. The MSHPs were installed to supplement (and partially displace) existing electric resistance heat; electricity savings were measured and analyzed. This program was also driven by the relatively low installed costs of MSHP in retrofits (~$3500–$5000 per head): they are much simpler to install in existing homes than a fully ducted gas furnace or heat pump system.

Measured coefficients of performance (COPs) for all installations averaged to 3; the results are shown with DOE climate zone and a representative city’s 99.6% heating design temperature in Table 1. During warmer parts of the heating season, COPs well in excess of 4 were measured. There is some relationship between measured COP and temperature conditions. However, the lower COP of the Inland Empire (Washington/Idaho) group was ascribed less to climate than to a preponderance of lower efficiency MSHP equipment.

Table 1. Measured MSHP Seasonal COPs with Climate Information (Baylon et al. 2012)

(St. Dev.)

DOE ClimateZone

99.6% Design
Willamette (OR)3.400.324C21.8°F
Puget Sound (WA)3.050.564C24.5°F
Inland Empire (WA/ID)2.410.595B2.9°F
Boise/Twin Falls (ID/WA)2.960.305B2.7°F
Eastern Idaho (ID)2.840.306B-4.9°F
Average Total3.000.55- 

The study demonstrated significant savings across the climate zones, relative to electric resistance heating. Cooling use was also monitored; MSHP efficiency was markedly better than the existing window air conditioner units, but overall cooling savings were small, compared to heating savings. This fact is easily explained by the small cooling load in this region, and the moderate increase in cooling efficiency (compared to the large increase in heating efficiency: COP ~1 for resistance heat versus ~3 for MSHPs). Surveys indicated that occupants were almost uniformly satisfied with the installed MSHPs.

In this study, the existing resistance heat was left in place (and operated at the homeowner’s discretion in bedrooms); the MSHP offset heating loads in the main living space of the house.

Another useful outcome of this work was that a large number of MSHPs were installed under field conditions. This resulted in a sampling of both high- and low-quality installations. The researchers collected and presented installation best practice recommendations for MSHPs, to obtain the best performance, reliability, and aesthetic results (Manclark and Thomas 2011).

2.1.3 Rosenbaum: Green Building Advisor Column

Rosenbaum (2014a) provided an overview of what he had learned using MSHPs in a variety of high performance projects. Advantages of these systems include elimination of combustion in the building (and venting), ability to supply both heating and cooling, and low installed cost. Lessons learned included:

Efficiency levels (coefficient of performance) are roughly in line with manufacturers’ specifications, based on comparing submetered electricity use to energy model predictions. However, these observations are not detailed efficiency measurements.

Low-temperature performance has been excellent in a variety of products and projects. Several MSHPs had sufficient capacity to meet set point below design temperatures, even without an intentional oversizing factor. Products designed for low-temperature performance (Mitsubishi “H2i” or “Hyper Heat” series) are rated to –13°F, and were still operating at –20°F. He reported that other practitioners have seen similar behavior out of units not actually rated for extremely low temperatures.

In a similar vein, Rosenbaum noted good output temperatures (120°F in December– January in zone 5A) in units designed for low-temperature operation, reducing the risks of cold blow complaints.

Temperature setbacks are not an effective strategy with MSHPs: when temperatures are set up, the unit runs at maximum capacity (and lowest efficiency) to return to set point temperature. This was primarily studied in terms of heating setback, but cooling setup may have similar behavior. In addition, these units are typically sized tightly relative to the load, and will therefore have longer recovery times.

Variable-speed cooling operation reduces the negative of oversizing (a common issue if the unit is sized for the heating load); anecdotal evidence suggests good dehumidification performance.

Rosenbaum has consistently measured carbon emission savings (and energy cost savings) when replacing existing fossil-fuel heating equipment (commonly boilers) with MSHPs, given the current fuel mix of the grid in the Northeast.

2.1.4 Rosenbaum: kWh/ft2∙Heating Degree Day 65°F Metric

Rosenbaum (2014b) presented similar material at the Northeast Sustainable Energy Association BuildingEnergy 2013 conference, covering the basics of MSHP systems and his project experience with this equipment. He presented monthly submetered electrical heating/cooling data, calculating the rough efficiency of several installations, . . .

Download complete report here.