Utilizing more efficient air conditioning equipment is an important strategy to curb residential energy consumption. The potential benefit from replacing an air conditioning unit with a more efficient one must be weighed against the burden associated with the creation of a new unit and disposal of the old unit. Using life cycle optimization methods, this research determines when a residential air conditioning unit should be replaced to minimize three separate objectives; (1) energy consumption, (2) greenhouse gas emissions and (3) cost to the consumer; over a fixed time horizon. Optimal replacement was explored over two periods, 1985-2025 and 2009-2025.
A model was created to determine the optimal replacement schedule for a central air conditioner (CAC) for each of the three objectives based on the climate where the unit operates. Six cities were considered: Ann Arbor, MI; Los Angeles, CA; Miami, FL; New York, NY; San Antonio, TX; and Wichita, KS. Life cycle profiles were developed for each model year CAC in the study. Dynamic parameters including CAC energy efficiency; energy intensity of raw materials and manufacturing; regional electricity grids characteristics including primary energy consumption, GHG emissions, and residential electricity price were included in the model. The production burdens were modeled using methods from both process-based life cycle analysis (LCA) and economic input-output life cycle analysis (EIO-LCA). A historical average efficiency improvement of 0.07 SEER per year was used to make efficiency projections into the future for a base case scenario. Other efficiency scenarios used the same improvement rate, but also modeled the adoption of a new federal efficiency standard at 15, 16, and 17 SEER in 2016. Finally, the consequences of poor installation, maintenance, and refrigerant recovery were explored in another set of scenarios.
Life Cycle Optimization Results
Using life cycle optimization (LCO) to calculate replacement schedules, it was found that energy minimization required more replacements than either GHG or cost minimization over the 1985-2025 time horizon. The energy minimization for the base case scenario required between 4 and 14 replacements after the initial CAC for the various cities examined. The EIO-LCA data for the production of a new CAC resulted in a significantly lower energy investment than the process-based LCA. The smaller the initial energy investment, the more beneficial it becomes to upgrade frequently to more efficient systems. Therefore, when EIO-LCA results were used in the LCO model, the analysis determined that more replacements (6-14 replacements) were necessary than when the process-based LCA results were used (4-10 replacements). For example, in the base case, the optimal replacement pattern to minimize energy in Ann Arbor, MI using EIO-LCA modeling of the production phase called for the original 1985 CAC to be replaced in 1989, and then replaced again in 1992, 1998, 2006, 2010 and 2018, but using process-based LCA results in the optimization resulted in replacements only occurring in 1992, 1998, 2006, and 2010.
When GHG emissions were minimized, 2 to 4 replacements were found to be optimal across all locations. For the locations examined, there were no differences in the number of replacements when using the process-based LCA or EIO-LCA data. In Ann Arbor, MI for the base case efficiency scenario, replacements were required in 1992 and 2010 using either process-based LCA or EIO-LCA data to model the production phase.
In terms of cost minimization under the base case, most locations required replacements in 1994 and 2007. Ann Arbor and Wichita required only one replacement in 2006 to minimize cost. This can be understood because Ann Arbor’s relatively low cooling demand and Wichita’s low electricity costs did not make it economical to upgrade more frequently.
The life cycle optimization was repeated while considering less ideal conditions where duct leakage was assumed and a lower 35% rate of refrigerant recovery was modeled at retirement instead of the base case assumption of 85% recovery. Also the level of maintenance was modeled separately using both a three service events scenario and a single service event scenario. In these service scenarios, the efficiency of the CAC was assumed to degrade between service events due to refrigerant leakage.
The optimal replacement schedules in the base case and these other less ideal scenarios were compared to other replacement schedules. A simple analysis was conducted to calculate the energy, GHG, and homeowner costs of replacing at a typical replacement interval of every 13-14 years and a maximum service life replacement interval of 20-21 years. In general, the environmental benefits of optimal replacement become larger when degradation is introduced and maintenance levels decline.