January 21st, 2023

Optimizing the Controls of a Rheem Heat Pump Water Heater

(Updated January 31, 2023 after a long series of experiments revealed surprising results)

This series of three articles is about my quest to optimize the energy efficiency, operating cost and indirect pollution effects of a new heat pump water heater using the controls in Rheem's EcoNet application for iOS or Android.  They allow one to choose between four operational modes (blendings of heat pump and electric resistance water heating) plus Vacation mode and Off.  They also allow one to set target temperatures for heating the water, and to change the mode and/or temperature up to four times per day, for each of the seven days of the week.  It took a long time to figure out a highly optimized schedule of settings and I'm hoping to make it easier for you to figure out this tricky and often misleading application's data so you too can get the most out of these amazingly efficient and low-polluting devices.

I’ve got a Rheem PROPH40 T2 RH375-15 40-gallon "hybrid" (heat pump & resistive elements) water heater that’s been operational for four months, replacing a 32-year old, 40-gallon Rheem natural gas unit.  As far as I'm concerned, this is the best water heater on the market, able to be both remarkably non-polluting (using only renewable power to run it) and remarkably economical (thanks to its extreme energy efficiency when used in its Heat Pump or Energy Saver modes).  If we lived in a much colder climate, the Japanese Sanden unit would be better, because it can operate down to about 20 below zero Fahrenheit with no resistive heating.  It’s in a garage, with air temps typically only a few degrees warmer than the outdoor temps when it’s cold and a few degrees cooler when it’s warm out.  The minimum operational temp of this unit's heat pump is said to be 35° F, but given our climate, I don’t expect it to fail to operate with excellent efficiency.  Surprisingly, subsequent testing over the 10 days since I first wrote this article has proven beyond a reasonable doubt that in our current conditions and with our low hot water usage Rheem's claim that a mixed use of the heat pump and at least the upper of the two resistive heating elements, allowed by choosing their "Energy Saver" mode, does actually use less power to get the job done than Heat Pump mode does!  It consistently used between 15 and 20% less power each day in Energy Saver mode than in Heat Pump mode!  It has four operational modes which allow your choice of allegedly all heat pump, all resistance, and two in-between modes, plus Vacation mode and Off.  According to the EPA, its actual water volume is 36 gallons, which I prefer to 40 gallons anyway as 36 gallons is still considerably more than we need.

Rheem claims that the “most efficient” mode of operation of the water heater is the “Energy Saver” mode, rather than the "Heat Pump" mode.  On the face of it, this claim appears absurd, given that the latter is described simply as a heat pump only mode and the former as a hybrid of both heat pump and resistance element operation.  Resistance heating elements can be pretty close to 100% efficient, but the heat pump can be close to 400% efficient, because it doesn’t simply convert electrical energy into heat, rather it uses that energy to move heat from the ambient air to the water in the tank.  Just like a refrigerator operating in reverse.  The yellow EPA energy usage stickers found on typical electric resistance water heaters foretell of them costing you over $500 a year for power at $0.12 per kWh.  The sticker on this one says the annual cost will be "$78" and I expect it to even cost less than any natural gas water heater to operate.  In fact given our climate and our hot water usage, I'm expecting that the power usage would translate to less than $60 a year were we to be paying $0.107 per kWh, as assumed in the EPA's Energy Star calculations.

In part three of this series of articles, I will explain the details of the experiments I ran to actually measure the power consumption of the two modes under nearly identical conditions of use, ambient temperature, and programmed schedule.

On-demand (tankless) water heaters cannot take advantage of the amazing efficiency of heat pumps and thus are either energy-hog electric models or polluting natural gas models, plus they are far more maintenance-intensive than tank types.  But they are compact, have very low mass for quake considerations, and have no standby losses from a tank.  If there's space for a tank and if it's well braced, by all means, please choose an electric heat pump model next time!  And be sure to operate it in its most energy-efficient mode, while also purchasing 100% renewable power from your power provider and/or making it yourself!  When methane leaks into the sky from throughout our natural gas systems, it warms up the planet about 120X more than the same mass of CO2 for so long as each gas remains in the sky.  It also diminishes to about half each 8+ years as it oxidizes to CO2.  So after about 27 years, leaked methane is down to about 10% of what it was initially and that strikes me as a reasonable point on the half-life type curve to generically describe its atmospheric lifetime with a single number.  In other words, the IPCC claim that the atmospheric lifetime of methane is 12 years is grossly unfair.  It's not that short.  It also takes 50 to 60 years to get to the point of being 98% gone.  CO2 lasts vastly longer, but about half of it goes into the oceans right away, in the first few decades (where it does other terrible harm aside from warming, making the water less alkaline and thus eventually dissolving away most living things in the sea).  The remaining CO2 in the sky lasts for as long as 45,000 years or as short as 500 years, depending on how you figure it.  But CO2 is basically forever, taking many thousands of years to finish turning into rock at the bottom of the ocean.  In any case, stop dumping your effluent into the sky.  It's fragile, or rather what lives under it is fragile.  Our civilization is actually a fantastically fragile system of thousands of interdependent parts, and most of them are going to wind up on the bottom of the ocean at the rate we're racing toward melting much of the land ice of the planet.  A hundred feet of sea level rise is by no means beyond our grasp.

The water heater can be controlled by an iOS or Android app, which lets you change the temperature target and operational mode up to four times, according to a schedule, on any given day of the week.  This gives us the chance to avoid using power during the peak pricing period of the day or to maximize the thermodynamic efficiency of the heat pump, or to use power when our photovoltaic system is producing more power than we’re using, while having perfectly tempered hot water at those times of day when we care to have it.  Pretty fancy.  The app also shows you exactly how much power it used and approximately when.  For each On cycle of heating, it reports the number of kilowatt hours (kWh) used, to the nearest one hundredth.  Note that Rheem actually has a version 6.x EcoNet app for controlling water heaters, which is presented in three different brandings; one for Rheem, one for Ruud and one that's somehow generic.  The one called simply “EcoNet” (in the App Store for the iOS version at least) is the one I've been using.

This is a fun puzzle, to try to optimize the minimization of power use and the minimization of power cost while still providing hot water at the ideal temperature (111° to 112° F for my taste) at those times when one might want to take a shower, and often without needing to mix any cold water in with the hot.  This means I’m aiming for the lowest possible operational temperature, rather than seeking to store more heat in the tank to obtain a larger effective capacity, for which I have no use.  Keeping the water temp at 111° F also optimizes safety by eliminating all scalding potential, unless, that is, the threat of dangerous bacteria actually breeding in the tank is real.  I'm skeptical that in a well-run water distribution system that pathogens will breed in the water distribution lines, at any temperature.

Since the first of this month (meaning since January 1, 2023), our power pricing has shifted so that the peak price is now between 4 PM and 9 PM.  It used to be in the afternoon, but the three big California utilities ganged up on us to convince the Calfornia Public Utilities Commission (CPUC) to shift the peak pricing for power to between 4 PM and 9 PM, cutting off rooftop solar power generation at the knees, more or less, with respect to keeping our power prices low and the price we’re given for the power we add back to the grid high.  For many years, our utility's highest power demand for the year as a whole was in the afternoon, on account of the large air conditioning demand from the extensive, warmer portions of PG&E's service area during the warmer months.  It's possible that so many of the homes that were paying through the nose for extremely high power usage have since gotten their own PV systems and thus cut this summer A/C use way down, but the utilities may also be exaggerating the 4 PM to 9 PM use to save themselves money dealing with we virtuous PV system owners who want to contribute to a less polluting grid.  In any case I want to at least keep the water heater off from 4 PM to 9 PM, 7 days a week.  There is no Part Peak power price for the E-TOU-C plan that we’re on.  It might be best for the grid though, if I were to turn the water heater back on at midnight instead of at 6:30 in the morning as shown in the first schedule below, because the grid is surely most flush with power between midnight and 3 AM or somewhat later, that being recommended by PG&E as the best time to charge an EV.  Conventional nukes are unable to throttle back their power generation to match the nightly lull in power demand.  Power plant operators want to be able to run their machines 24/7 to make more money from their large capital investment.  On occasion, nuke operators have been forced to pay people to take their power (negative price).  Theoretical, future, molten salt LFTR reactors will have superior load-following capability, as well as being far superior to conventional nukes in just about every other way — if their engineering issues can be worked out, especially corrosion problems with the FLIBE coolant.  But for now, renewables are the thing, and thank goodness their cost continues to plummet.  It's time to build a better grid.

We don’t have a dishwasher and if we did I wouldn’t want it to use hot water anyway and certainly not water heated extra, hotter than the hot water heater would provide, just for alleged cleaning and sterilization advantages.  Nor do we use a drop of hot water for laundry (the hot water hose is disconnected entirely and a telltale water pressure gauge is attached to the hot water faucet of the laundry sink, where it gives valuable, routine reports of both current and max water pressure.  It shows impressively effective dampening of our closed system pressure from water heating cycles by the 2-gallon Watts pressure tank with 50-psi pressurized air bladder and with our pressure reducing valve for the street water at 120 to 130 psi set to 50 psi for the house.

And I want to optimize heating efficiency by heating when ambient garage air (quite chilly in there these days, around 40 to 55 F) is warmer, if possible.

And I want to minimize standby losses by having lower tank temps when hot water is of no use to us and when ambient (garage) air temps are lower (if possible).

As well as wanting to use some of our own power generation, when we’re making more with our 3 kW PV system than we are using (mid-day, early afternoon).  This keeps us more independent of the grid and helps prevent the economic deal of the power market from disfavoring us.  And yes, rooftop and utility-scale solar still should be subsidized so as to most rapidly drive our abandonment of fossil fuels.  We are badly losing the race to de-carbonize our energy systems in time to avoid unspeakably immense and tragic damages which are plainly in our future.  My #1 issue among the list of climate catastrophes is the sea level rise which many scientists already know perfectly well is going to drown most of the world's coastal regions to increasingly catastrophic degree as we put a large fraction of the world's land ice back into the ocean where it used to be.  So take fossil fuels out of your future and pray that cheap carbon capture and sequestration can be engineered and done on a massive scale, before we lose most of humanity's prime habitat forever to the curse of salt water rising many tens of feet.

So far, my only two measurements of the rate of water temp increase in the usual, hyper-efficient Heat Pump mode have been 3+° F in a half hour and 3° F in an hour, both surprisingly (if not distressingly) low.  I suspect it’s actually more like 5 or 6 degrees per hour, so more measurements of that are needed.  To date (as of the original date of this article), 100% of operation has been Heat Pump (allegedly meaning heat pump only) mode since it was installed, despite Rheem’s constant, hard-to-believe badgering to use “Energy Saver” mode to get the best “efficiency."  And with the nominally 40-gallon water heater (EPA volume is 36 gallons) simply set to operate to 111° F at all times there has never been any clear suggestion of problematically slow recovery times, plus our max need for hot water at 111° F during any one hour is only about 12 gallons, needing 5.5 gallons for one shower with a good quality shower head that doesn't waste water.  111° is the temp that yields water needing little or no tempering with cold for a shower.  Our hot water lines are insulated half-inch copper and the shower is about 50 feet from the water heater and we do use a recirculating loop and AquaMotion pump system which is activated by a radio signal and takes about 90 seconds to get fully heated water to the bathroom in question.  [See my upcoming Part 3 article for the results of actual testing of Heat Pump mode vs Energy Saver mode.]

The free Rheem app's new version 6 with the mostly white interface is much better than the old version 5 with the dark interface.  It still has a bunch of issues about the way it reports power usage and is short on some features, but otherwise works very well.  It has taken a lot of figuring to understand just what its energy usage data really shows (see below).  Here is my first test schedule.  I've made many changes since.

Note that a max of four schedule changes are provided per day, but I’m only using three here, as 9 PM is no change.  I’ve chosen to leave the unit essentially off by using Vacation mode from 4 PM through 6:30 AM [later it has turned out that off isn't really off].  Since you’d expect the heater to turn on to warm itself back up starting at 6:30 AM and perhaps be ready for a 111° F shower by 8:00 AM, this cycle will mean that the heat pump is working a lot at a point where the garage is apt to be it’s coldest, but our weather report for this week indicates very little cooling from say midnight until 6 AM, so that's not much less efficient than heating just after midnight would be.  Heating extra before 4 PM is another possibility, and which is best depends on just how well the tank's insulation manages to avoid standby losses.  Each time one uses a recirculating loop or runs hot water from the tank to a faucet, that water will loose its heat fairly quickly, so one use of our recirculating pump costs us the heat of at least one gallon of water, the volume of the loop.  The pump is on for under 5 minutes each cycle, after we activate it with a button press on a remote control in the bathroom.  This can be done many ways, and for a long time we used a recirculating pump that was on a timer, turned on for two 4 hour periods each day.  That led to much greater standby losses from the insulated pipes of the 90+ foot loop than our new pump setup, purchased together with the new water heater.  Our water has low CaCO3 equivalent and so is easy on water heaters, and we always kept the water temp on the low side, which also helps, but Rheem still did very well for that gas heater to last 32 years with only one incident of maintenance, before it started slowly leaking from the bottom.  Over three months the leak didn't become a flood, thankfully, giving me time to plan the big conversion process to the new water heater, which was complicated in ways that it would not be for most situations.  Still, there was a new circuit, and I had to learn a lot, etc., etc., and overall it was a big project.  Replacing the current unit should be vastly easier when it finally wears out.

I thought it might be better to use Off mode than Vacation mode [it turned out to make zero difference].  The latter supposedly allows the water to cool down to as low as 65° F but no cooler.  I’m guessing, based on a measurement at the kitchen sink, that ours is cooling from about 111° to about 100° F in the ~15-hour overnight period [As of ten days later I have concluded that the tank loses about 0.7° F of heat per hour, starting at around 115° F from standby losses with the ambient temp averaging about 44° F].  Note that the temperature targets allowed by the EcoNet app only go as low as 110° F but this should not prevent eventual cooling to as low as 65° F in Vacation mode.  It would take a few days of being in Vacation mode or Off altogether to cool down that much.

Rather than simply set the water heater to run in Heat Pump (only) mode from 6:30 AM to 4 PM, I decided to also goose the temp to 115° F starting at 3 PM so as to ensure sufficient heat for a shower in the late evening, despite the water heater being essentially off starting at 4 PM.  This also has some advantage from the ambient garage air temp being higher for more efficient heating as well as for better utilization our PV power.  In other words, store heat when its best to do so.

So in the Schedule screenshot above, think of the actual temp targets for the Vacation mode period as being 65° F, not 110° F.

Now on to the observed electricity use for the first 20½ hours of today and all of yesterday.  Yesterday’s pattern may have been the result of a somewhat different schedule during the earlier part of the day, so today’s results (the bars) are the ones to consider valid results.

First thing: Rheem’s plotting of consumption seems to be consistently in error with respect to the time of consumption shown.  For example they show 0.47 kWh of use at the 9th hour of the day, i.e. the one hour period ending at 9 AM, but this was already being shown shortly after 8 AM and the compressor was already off then and has remained off for the next hour and longer, so this datum actually must be for the hour ending at 8 AM or possibly even earlier.  I'm pretty sure the rate of power consumption in heat pump mode isn't high enough for it to have consumed a half a kWh in the first 20 minutes or so of the 8 AM hour.  Thus it is probably an accounting of energy consumption of the 8th hour, and shifted misleadingly one step to the right in the graph, almost certainly along with all of the other data.  But I'll try to verify that later.  (Better insights into this below)

Second thing: We can see uses of 0.10 and 0.11 kWh’s at the 1st and 5th hour positions.  Why does the water heater use power when in Vacation mode with a tank full of water that is still about 40° F hotter than the set point for Vacation mode?  Could it be that the system is trying to maintain 110° F after all?  If so, the usage seems inadequate to that task.  One YouTube video maker foolishly ascribed a considerable amount of power consumption to the operation of the WiFi features of the water heater, which is surely off target by roughly a factor of 100.  I.e. the electronics consume almost nothing, so that isn’t the answer here, and the timing pattern would make no sense either.

Third thing: Why would the water heater wait until the hour ending at 8 AM to turn on so as to reach the 111° F target, when it’s set to renew seeking that target temp starting at 6:30 AM? One would expect two bars at least: one for the hour ending at 7 AM and one for the hour ending at 8 AM, if not also a third one for the next hour, assuming the recovery were still incomplete at 8 AM.  (Answer below)

Is it possible, via some weird physics voodoo, that rather than the whole, fully heated tank cooling down pretty uniformly while it’s in Vacation mode that water at the top of the tank cools down a lot less, so the water heater logic algorithm says “I’m OK for a shower despite having sat here in this cold garage for 14½ hours in Vacation mode”?  I can’t imagine this is possible.  What on Earth are those short heating cycles for during the night?

It is also possible that the objective of simultaneously minimizing power consumption and total power cost can be met almost precisely as well and conceivably even a bit better by only using Vacation mode (or Off mode) between 4 and 9 PM.  Plus the power cost may well have three tiers, not two: peak, partial peak and off peak.  But after double-checking it turns out that our new pricing plan only has two tiers.  That would likely argue for keeping the water heater on all night, starting at11 PM or perhaps at midnight, or whenever off peak pricing would start or when the grid needs help by using power when it's most abundant, starting at around midnight or a little later.  Off peak pricing or a desire to help the grid would also argue for starting Vacation mode earlier than 4 PM.

We are fortunate here to have the option to pay $0.01 more per kWh for 100% clean, renewable grid power over the price for the roughly 60% clean power that we would otherwise get, purchasing power from MCE instead of from PG&E.  The latter delivers the power and handles most of the billing process.  Delivery (the cost of the grid) is a few times greater than the cost of the power, which is why the utilities (principally PG&E, Southern California Edison and San Diego Gas and Electric) hate to see rooftop solar get to avoid paying for power, because that payment covers both the grid cost and the power cost.  Solar needs and deserves considerable subsidy, because it makes future life better for everyone and drives future low solar prices for everyone, despite the fact that it can already save many people money over the long term.

We have been generating over 90% of our own power for over 16 years now, with a nearly 3 kW PV system and purchasing 100% clean power for the little remaining amount starting a few years ago.  By switching to heat pump electric water heating from gas and to a beautiful Miele heat pump clothes dryer from a conventional electric clothes dryer, we will probably wind up still using under 15 kWh per day instead of our prior average of a little under 14 kWh, on an average annual basis.  When I have to replace my trusty second-gen Prius, the EV will increase our power use at lot, of course, so being able to buy clean power from the grid and using power at preferred times of day for grid health will become even more important.  We also expect to eventually replace our small, 75-year old gas space heater with a heat pump system, for both heat and even a bit of cooling here and there during hot spells (which would unfortunately be when the grid is most overloaded).

Before we installed our PV system, during a re-roofing, we were using more like 21 kWh a day but I cut that down by a third via various efficiency improvements.  The average American home as of about 2007 was using around 27 kWh a day.  The lower we can get that, the better everything will work.  Then again it has to go up as we stop using natural gas and heating oil and gasoline as fast as we possibly can.  Gone are the days when bringing a new gasoline-powered car into the world is a reasonable thing to do.  Hold out for a mature EV.

—JH

This evening I switched the schedule so the Vacation mode times became Off times instead.  Tomorrow’s power use data will show the effect of that [It made no difference].  I expect three of the small, i.e. one-tenth of a kWh use periods will disappear (they didn't!).  So far, daily power use since September has been roughly in the range of 1.15 to 1.6 kWh, the former being in September and the latter being during the much colder month of December, on account of both reduced heat pump efficiency trying to pull heat out of the colder air and greater standby losses from the tank, with the colder garage often being in the low 40’s.  September’s average temps in the garage were probably in the 60’s.  Our average use of hot water per day (meaning total water volume that flows out of the tank itself and isn't returned still hot) is probably only about seven gallons.

This schedule is still pretty different from the one I ultimately settled on.  See my Part Three article.

Update 7:25 AM January 21st, 2023:  After being in the Off state overnight (or so I thought), the exact same two small usage periods had occurred as of 7 AM, in the 1st and 5th hours, and none through the 8th hour (yet), as occurred in the Vacation mode!  Also, at 7:15 AM the water temp leveled off at the kitchen sink at 110° F (taking over two minutes of flow to warm up the last little bit), with the garage air temp at the top of the water heater where air is pulled into the heat pump being just under 40° F.  Following the schedule, the unit went from Off mode (turns out it was probably in Vacation mode after all) to Heat Pump mode at 6:30 AM, so however much the tank temp fell in the fairly cold garage since being set to at first Vacation mode starting at 4 PM then switching to Off mode at around 9 PM, the water at the top of the tank was all the way to within one degree of the target temp after only about 45 minutes of operation.  (Later I saw that the heating was completed before 8:20 AM and a usage spike of 0.50 kWh appeared for the 9th hour.  That means that heating did continue beyond 8 AM and that, sure enough, the usages reported in the Energy Consumption panel are usages that ended in the hour shown.  Thus a bar in the graph at "9" means, in this case, 6:30 to about 8:10, i.e. power used in three different one-hour periods.)  The EcoNet app seemingly should have shown power usage for the 7th hour as of 7:00 AM or at least by 7:01 AM with a screen refresh, but it did not.  Perhaps all energy use periods remain unreported until the heating cycle ends (Later: indeed that seems to be the case).  And even then, they might be further delayed until the end of the last hour within which they occurred (That apparently does not occur, so one would surmise that the water heater's control circuitry doesn't allow a second cycle to occur as soon as 60 minutes after the end of another cycle, however if a cycle were to end at 1 minute past the hour, and then a lot of hot water were used, you would think the water heater would immediately start to heat more water.  The usage reporting problem this would create could be solved by changing the reported use amount for that hour, assuming the fresh usage period ends within that same hour.).  That would make sense.  Nevertheless the 0.10 to 0.11 kWh usages that seem to occur on a once each 4 hours schedule in Vacation mode still occurred during this initial test of Off mode.  This was surprising (later I found this to likely be an error - the water heater was probably still in Vacation mode, so this morning's data merely served to show essentially perfect consistency of behavior with yesterday morning's behavior.) (However, a further test the next morning did show that these odd usages also occurred when set to Off after all).

Further Update!  7:35 AM January 21st 2023:

To continue, I'm trying to figure out exactly what's going on in yesterday's Energy Consumption panel, shown above.  Notice the marked tendency of the unit to exhibit usage in the first hour of each four hour period.  All six such hours showed usage.  Two of them were higher usage amounts.  I know why the spike for the 17th period occurred but the spike of the 9th period is a mystery.  It looks like a spike from hot water usage, but we're not aware of any usage having happened.  Maybe it was just a little at the kitchen sink.  Also, the spike for the 15th hour, the only usage to occur in a period which was not one of the six first-of-four periods, appears to be a usage spike, but again, we're not aware of any usage between 2 and 3 PM.  I'm going to have to pay closer attention to that in order to figure out exactly how a new cycle is stimulated.

In order to get this right you've got to know just how fast the tank cools down.  If I measure the water temp at the kitchen sink at 6:20 tomorrow morning, before the water heater wakes up, I'll have a much better idea of that.  I'm starting to think the heat loss is under 1 degree per hour, but still likely over a half degree [again, I later settled on 0.7 degrees per hour of standby loss].  We really should try to heat our water mostly when we're generating more PV power than we're otherwise using, which happens nearly every day, but not on stormy winter days, which are too dark.  I'd guess that on average, our highest power generation occurs between roughly 11 AM and 3:30 PM standard time.

I will soon post a second article (nearly finished) with more conclusive experimental results and a fairly detailed explanation of how the EcoNet app works.