The Home Automatic Utility System project was begun as a contest entry. The Plat'Home Company was looking for innovative uses for their "Open Micro Server" product, a compact and robust server computer, based on the RMI Alchemy Au1550 400 MHz MIPS-architecture chip, running Linux. (See information on contest winners.)
The HAUS system, as it has developed to date, operates continuously to take data from our oil-fired "hydronic" heating and domestic hot water system, including status of the burner, circulation pumps, hot water outlet temperature, and outdoor ambient temperature. Readings are taken every 30 seconds, stored in Compact Flash media (solid state disk for the OMS), processed, and made available as daily reports via the Apache web server. The system operates autonomously and reliably, communication with the household LAN via a WiFi adaptor.
Note added April, 2013. The HAUS system was retired when we converted to a high-efficiency natural gas boiler. The new system comes with microcomputer control, but there is no interface for external digital logging or control. (Yet!)
2 Results and Analysis
Typical daily plot on a cold day:
The top graph is a temperature measured at the domestic hot water (DHW) tank outlet; it indicates hot water usage. The middle plot is outdoor ambient temperature. The graph below that shows on-off values (bottom up) for (1, red) domestic hot water circulator; (2, green) burner on; (3, blue) zone 1 heat demand; (4, purple) zone 2; (5, cyan) zone 3.
A plot of minutes of furnace burn time vs average daily ambient temperature, with straight line least squares fit, for 46 days (5 October through 23 November, 2008):
I note that since undertaking this project, we have kept our household much colder than normal, with thermostats turned up only when someone complains. This is a benefit of getting daily feedback on fuel costs!
2.1 Domestic Hot Water
One point to note is that for our household (two adults), the percentage of burn time attributed to DHW needs is modest. Circulator on time averages 44 minutes/day, which causes less than 20 minutes of burn time. In the "idle" condition (summer weather, low DHW demand), the DHW load accounts for perhaps 1/3 of the burn time. In "fall" conditions, we have heat demand averaging around 90 minutes of burn time, so DHW averages no more than about 20% of oil consumption.
We could consider further work on the DHW system - more insulation, lower temperature operation, etc., but the conclusion is that there is not very much to be gained from this. A 10% reduction in burn required for DHW would be fairly hard to achieve and would produce only a 2% savings in our energy bill.
2.2 The Summer Problem
Note that burn time appears to pass through zero at about 19C. This is an artifact of analyzing only "cold" days on this graph. Unfortunately, the hot water system needs to burn around 20 minutes per day just to sustain itself at operating temperature without any heat or DHW demand.
Heating experts tell me that it is "dangerous" to allow the boiler to cool down in summer time because of corrosion (rust from condensation, presumably) and mechanical stress (from cooling to ambient). The latter seems dubious to me, if we are only considering one cool-down per summer season.
2.2.1 Reducing Idle Demand
Can we avoid the expense of $1-$3 per day for "idling"? This would appear to be a summertime issue, but actually the cost represents heat that is lost in winter also, except it is lost in the larger demands for "useful" home heating.
- Changing furnace set-points. Some preliminary work indicates that there may be some gain from reducing the temperature of the lower limit point (at which the burner turns on) to perhaps 60 C or lower. This will reduce the time average boiler water temperature, producing some small savings. In addition, there is some suggestion that reducing the number of burn cycles reduces the total burn time. That is, a few longer burns cost less than numerous short burns. Furnace information sites (ref?) suggest that there is a kind of "set up" overhead to bring the burner into efficient operation at the beginning of a cycle. It is possible that increasing the upper limit point (when the burner shuts off) might actually save fuel.
The limit on the lower limit point is set by the DHW heat exchanger, which needs a boiler hot water supply substantially above the DHW set point (40 C) in order to recharge quickly.
- Reducing heat losses. The burn time in idle conditions is determined by the heat loss from the furnace system. Temperature measurements can establish where the largest losses are, suggesting where to install new insulation.
- Boiler shutdown. Summer shutdown should not be impossible, despite advice from the "pros". The main thing is to keep all metalwork above the local dewpoint temperature to eliminate condensation. A combination of electric heating and/or basement dehumidification should work, but these measures need to be reliable -- and less expensive than the idle fuel costs!
The main problem with a shutdown is DHW service. We would need to install an alternative hot water heater. Would the capital cost be justified?
Software for this project was implemented in Python 2.4, supplemented with gnuplot. These routines are provided "as is" (no support!) as Open Source Software under the General Public License (GPL) v. 3. Programs include:
- ulog.py - A program that loops "forever", coming to life every 30 seconds to take readings and store results. At 0 hours local time, the current log file is closed and the next day's begins.
- udaily.py - Run as a cron job shortly after midnight to process the past day's log and generate a report containing a text summary and graphical display.
- gnuplot.prog - The gnuplot program launched by udaily.py.
- uweb.py - A program to generate HTML files for web-based reports.
- udaily_csv.py - A special version of udaily.py that scans all old logs to produce a CSV format text file that can be used for off-line statistical analysis.
The installation can be seen in the following photos. (Click for larger views.)
In Figure 1, we see the domestic hot water (DHW) heat exchanger at left. Its circulator pump is at floor level. It is set to maintain a temperature of ~40 °C. A DS18B20 temperature sensor is installed in contact with the DHW outlet at the top of the tank, under the pipe insulation. The cubical box in the center is the oil-fired furnace (or "boiler") which maintains its water temperature at 70 - 80 °C. Most of the controls are on the wall at the rear, along with the computer components. The three zone circulation pumps for heating are visible on the wall behind the furnace.
Figure 2 shows a detail view of the control wall. The white box at top is an Encore ENRXWI-G configured as a WiFi bridge to the household WLAN. Below and to the right a black box which is the Plat'Home OMS-AL400/128 computer. Just below is the DLP-IO8-G Data Acquisition Module mounted on perf-board. The 3 gray boxes are the pre-existing zone circulator controls. One of them provides a spare set of relay contacts, but two required addition of a circulator interface (below). The small black block at the left of the panel is a furance controller interface, built around a 120 VAC to 12 VDC power block ("wall wart").
The thermostatic control for the DHW circulator has a spare set of relay contacts, as does one of the 3 zone controllers. These were simple to interface to the DLP-IO8-G, with a simple pull-up resistor. The furnace controller provides a switched 120 VAC power signal, which must be converted safely to a contact closure. The surplus wall wart DC converter provides a simple and safe power conversion, followed by a simple reed relay circuit:
The zone controllers needed a similar interface to translate 24 VAC to a contact closure. (All parts are available at Radio Shack.)
Temperature sensors are implemented with the Dallas / Maxim DS18B20 Programmable Resolution 1-Wire Digital Thermometer, which are available in a TO-92 package. The DLP-IO8-G Module has a temperature mode designed to operate with this device. Conversion time is slow (750 mS), but adequate for our work. The thermometer chips are connected according to instructions provided in the DLP manual.
Updated: 24 Nov. 2008, Wikified: 3 Jul 2010