Written on: April 18, 2022 by Phillip J. Baratz
With all due respect to REO Speedwagon, the band’s 1981 #1 hit “Keep On Loving You” pops into my head every time I hear a client tell me about the time and effort they spend adjusting (just a few) Ks (K-factors, the calculations that show how quickly a home is consuming the fuel it uses to produce heat). We have all faced times where the Ks just don’t look right, and in many cases, we just don’t want to wait for the back office system’s (BOS) correct formula to catch up with the actual heating degree day (HDD) vs. consumption calculation.
What if the calculation wasn’t the problem?
Allow me to paint a picture for you with two examples based on weather and associated consumption for two sets of your customers:
Scenario #1: You have customers at 1 Main Street, 2 Main Street and 3 Main Street
You fill all three customers’ tanks on Feb. 1 and again on Feb. 22, each fill is 201 gallons. During this time period there were 806 HDDs.
On Mar. 27, each of the three customers receives another delivery (of different amounts).
• 1 Main Street: 202 gallons
• 2 Main Street: 135 gallons
• 3 Main Street: 98 gallons
During this time period (Feb. 23–Mar. 27), there were 809 HDDs.
Scenario #2: You have additional customers at 4 Main Street, 5 Main Street and 6 Main Street
You fill all three customers’ tanks on Jul. 1 and again on Dec. 22, each fill is 200 gallons.
During this time period (Jul. 1–Dec. 22), there were 1,596 HDDs.
On Jan. 17, each of the three customers receives another delivery (of different amounts).
• 4 Main Street: 99 gallons
• 5 Main Street: 132 gallons
• 6 Main Street: 198 gallons
During this time period (Dec. 23–Jan. 17), there were 791 HDDs
What happened & what changes would you make?
The problem with K-factor accounting (or consumption) is not that the homeowners are misbehaving. It is not that they suddenly expanded their houses or went to Florida for a few months. It is not that the BOS doesn’t know how to do math.
The problem (and it seems to apply to most customers) is that the notion “x HDDs = y gallons of consumption” is only true in the middle of the winter. If the heating system is on for the entire period in between deliveries, then a “true K” can be calculated. However, once you peek outside of the middle of the winter and include additional days in your calculations, all bets are off.
Let’s shed light on the “mysteries” above. Scenario #1 looks like three houses with four Ks, 806 HDDs and 200 gallons of consumption—this is pretty straightforward. What happened after the next 800 HDDs? Shouldn’t there have been ~200 more gallons of consumption? Maybe the customers didn’t really consume a gallon for every four HDDs? The answer is that, indeed, these customers do consume one gallon each time there are four HDDs. They have a (usage) K of four. However, while Customer #1 kept the heat “on” all the way through the delivery date of Mar. 27, Customer #2 shut the heat off on Mar. 14 (the first warm day of early Spring) and the HDDs from Mar. 14–27 didn’t cause any fuel consumption. Furthermore, Customer #3 followed the path of Customer #2, but took a vacation starting Mar. 7 and only consumed fuel until that date, resulting in consumption that was limited to ~100 gallons.
The issue was not the K, it was whether the heating system was on
Scenario #2 appears to any outsider as a customer with a very high K. It took 1,596 HDDs to consume
~200 gallons. Lacking any other information, this looks like a K-factor of eight. Why then did the next delivery to the three customers look so different from each other? Is the K of these neighbors really an “8”?
The answer is almost the polar opposite (pun intended) of Scenario #1. Customer #4 had the heat on during the entire period from Jul.1–Dec. 22 and consumed 200 gallons. From Dec. 23–Jan. 17, the 791 HDDs caused consumption of 99 gallons, due to their K of eight. Customer #5 has a K of six but didn’t turn the heating system on for the winter until Oct. 31, consuming 200 gallons from then until Dec. 22. From Dec. 23 through Jan. 17, Customer# 5’s K of six resulted in consumption of 132 gallons. Customer #6 put on the heat around Thanksgiving (Nov. 24) and has a K of four, resulting in a 200-gallon delivery on Dec. 22 and a great Jan. 17 delivery of 198 gallons.
While hindsight is 20/20, we need a way to forecast tank levels, not a way to backtrack and figure out why things didn’t turn out as expected. Each time a delivery is made, the BOS uses its rearview mirror and adjusts. The adjustments are logical, and they are conservative—as they should be. If consumption exceeded expectations, the K will drop (by a formulaic amount). If consumption was less than expected, which is a much more frequent occurrence, the K will rise, but disproportionately from the lowering of the K (to avoid the increased likelihood of a future runout).
If you, as an owner or a dispatcher, are not pleased with the tempo of the “formulaic changes of the K,” you would join the club of “K changers” (you know who you are!) and would manually make some changes—almost always more aggressively than the BOS formula, and often an overreaction. For those interested in the psychology of manual changes vs. formulaic changes, look up the term “Recency Bias”—it will explain a lot about human nature!
Are remote monitors the only solution?
Simply put, each day a monitor will report how much fuel is in the tank. Armed with that information, and the proper delivery planning, you should avoid delivery surprises. Take note that not every 200-gallon delivery is optimal. There is a lot more to “optimal” than simply the size of individual deliveries, but we will leave that discussion for another time. The purpose of the monitor is to know the size of the delivery you will be making.
However, monitors cost money. In most cases there is an ongoing monitoring fee (ADEPT alert: we have a program that waives 100% of monitoring fees). Monitors take time to install and not every monitor reports every single day. If monitors, installation and monitoring were free, you would have them on 100% of your tanks; if they cost thousands of dollars for a residential tank, you would have them on 0% of your tanks. Cost effectiveness is directly related to the benefits that can be achieved—not only the “knowing,” but the value of knowing. Better put, monitors should be on the tanks where the delivery size is unknown. For tanks where you do know the delivery size, and there are many of them, you may be better off without a monitor.
How do you know which is which? As always, the answer is in your data. Analyzing your deliveries can be a manual process or an automated process (like many other things). It can be based on averages, histograms, standard deviations, etc. We prefer to point out extremes after we analyze data. We start at opposite ends of the spectrum (I was reluctant to use the term “polar opposites” again). We would tell you which tanks need monitors the most and which need them the least. Working from both ends towards the middle has the effect of driving your “average predictability” up in an accelerated manner. If you find your biggest outliers (formulaically), each tank to which you add a monitor will provide a bigger benefit than the one that follows.
Summary of key points:
• Everyone hates run-outs
• Homeowners turn their heating systems “on” and “off” in unpredictable ways
• Deliveries are generally smaller than anticipated
• Ks adjust conservatively by their formulaic nature
• Manual adjustments are often overreactions
• Monitors are not needed on all tanks
• Selection of which tanks benefit the most from monitors can be automated. ICM
