Data Loggers:

An Interview with Some Heavy Users

Home Energy spoke with four experienced data logger users. Each of our interviewees learned their lessons the hard way. Here, they offer tips and discuss pitfalls.

The first data loggers were human.  When labor was cheap, students and others were hired to sit next to wattmeters and record the reading every 15 minutes. Data loss occurred for various reasons, such as boredom.

Digital data loggers have been refined into valuable energy-auditing tools. They make it easy to collect detailed information over a long span of time with a high degree of accuracy. But many auditors and contractors are still hesitant to use these expensive, high-tech devices. They still depend on instincts, senses, and instruments that give one bit of data for one moment in time.

We think data loggers could be used more widely by auditors, HVAC contractors, utility staff dealing with high bill complaints, and others in the energy and comfort fields. We assembled a group of four experts to describe how they use these tools---what they've learned, and what they regret. Their experiences might suggest a way that data logging could help you to monitor buildings or equipment.

Danny Parker is principal research scientist at the Florida Solar Energy Center. His most recent Home Energy article covered an intensive data-logging experiment on a home in Florida ("Florida House Aglow with Lighting Retrofit," Jan/Feb '97, p. 21). Fredric Goldner is principal at Energy Management and Research Associates in Brooklyn, New York. In the July/Aug '96 Home Energy, he wrote "Try These On for Size: New Guidelines for Multifamily Water Heating" (p. 32). This article describes how his company monitored 30 multifamily buildings in New York. Jonathan Beers works for Madison Gas and Electric, which helped create a program to lend portable energy meters to customers directly through Madison-area libraries. He wrote "Condensing Furnaces: Lessons from a Utility" for the Nov/Dec '94 Home Energy (p. 13). David Springer is President of Davis Energy Group, which has completed numerous monitoring projects for utilities. Home Energy Executive Editor Alan Meier spoke with these four experts via teleconference.

Home Energy: What sort of loggers do you use, and what do you use them for?

David Springer: We use Data Electronics models DT50, DT100, and DT500 series (made in Australia). Also, Synergistic Control Systems, Campbell Scientific CR10, Fluke, and ACR Smart Reader. One of our largest residential sites was in Palm Desert, where we monitored 56 data points at one house, including temperature, water and gas flow, power, solar radiation, and wind.

Danny Parker: We use them to monitor energy end uses within residential and commercial buildings as well as meteorological conditions. From a typical building we usually gather between 20 and 30 channels worth of data.

Other than Campbell multichannel data loggers, we use some of the Pacific Science and Technology motor loggers and light loggers, to pick up lighting and other individual devices that otherwise are difficult to monitor.

Fredric Goldner: For the last six years, we've collected data on heat and domestic hot-water use in multifamily buildings. Monitoring allows us to model using real-time flows and temperatures in systems instead of using theoretical consumptions. We have collected data on fuel consumption, weather, indoor air quality, boiler cycling, apartment temperature, domestic hot water temperature and flows, boiler stack temperature, and makeup water. This was then analyzed. We came up with new system sizing and design methods and determined building dynamics. Most of my projects have had a couple dozen data points; some of them have been down to as few as four channels in a building.

Jonathan Beers: In the residential part of the company, we mostly use portable kWh meters to measure appliance energy consumption. I've mainly metered fridges in low-income households to find guzzlers that are candidates for replacement. The results surprised us. We found only a loose correlation between electricity usage and labeled amp draw, age, or even side-by-side versus top freezer styles. We've used three different brands of meters. As part of a grant from the Wisconsin Environmental Education Board, we put portable energy meters in Dane County libraries, so people can check them out using library cards. In-house, our meter readers have been loaning out portable meters for the last 15 years or so.

HE: Do customers like them ? And what do they use them for?

JB: There's a tremendous waiting list for the ones at the library. Mostly, they're using them to help decide whether to replace refrigerators. One guy was trying to figure out phantom loads--what used electricity even when switched off. They can measure anything that plugs in, at least if it's 110 volts. The meter readers have a couple of meters for electric clothes dryers, but they don't work on other 220 volt appliances, which have a different plug shape.

HE: What do you use?

JB: In-house we have some ancient Landis and Gyr. The ones in the libraries are Line Loggers from Pacific Science and Technology. We did a neighborhood-based program where the contractor chose to use the Energy Tellers, because they have the lowest cost. They weren't nearly as accurate, though. You get what you pay for.

DP: We've used different types of loggers. We've had the best luck with Campbell data loggers, which we now use almost exclusively. Even though their learning curve is steep, for a multichannel datalogger, they are extremely reliable. Reliability is really important.

DS: We've settled on Data Electronics for monitoring temperature. It's a really good analog input logger. For power monitoring, we prefer Synergistic Control Systems. The Data Electronics requires a separate power monitor to provide a pulsed input to a counter; with the Synergistic Control Systems you can connect a current transformer directly to the logger.

FG: Our greatest success has been with loggers that are integrated with existing building control systems. I take probably 80% of my data with OAS Heat Computers. Those act as the building boiler and temperature controller in multifamily buildings. These really extensive research projects would have been prohibitively expensive if we were not hooking into existing building systems. Controllers cost $5,000- $10,000 a pop, but for about $1,200 extra in each location, we had the company add memory board upgrades and a number of additional data input points.

HE: Danny, have you ever tapped into the HVA C thermostat for a single family building? (See "'Read Me Your Thermostat'.' Short-Term Evaluation Tools," Home Energy, Mar/Apr '93, p. 31)

DP: No, but we've tapped into the pulse-initiating meters that utilities have installed. We generally measure power with a single voltage tap off the main electrical panel and use split-core current transformers that go around the individual leads right there. We usually locate the data logger right there. To measure interior temperature, which is critical for cooling, we put a type-T thermocouple exactly where the thermostat is. We have people who are very clever at doing that so it's difficult to see it's there. You need craftspeople to install data loggers. They have to do nice looking work, so people don't see this as any kind of infringement on the beauty of their home or building.

HE: Have loggers ever answered questions you couldn't have answered any other way ? 

DP: All the time.

FG: In our first DHW research project, we collected 15-minute flow and temperature data for 14 months in 30 buildings, totaling over 12 million data points. There's no way we could have even considered doing that without loggers. I read a paper done by AGA Labs back in 1904 or something, where they looked at essentially the same parameters as we did, but they just had one gentleman sitting down in the basement, reading the meter for three days.

DS: We're currently monitoring geothermal heat pump systems. We couldn't accomplish what we're doing--taking data and calibrating computer models with that data--without being able to take short-interval, long-term data.

HE: Have you ever thought you knew what was happening with a house, only to have monitoring show that you were wrong?

DP: All the time.

JB: I've metered round-top, single-door fridges from the '50s where the doors wouldn't even close. But they're still not guzzlers, because the freezer section is small and is barely cold enough to make ice, and there's no automatic defrost. So they don't use much, even with a door that won't close right.

DS: You find things you don't expect, and those are sometimes the most important. We plotted a curve of coefficient of performance (COP) versus water temperature for one of the hydronic heat pump systems we're monitoring. We found that the COP at 140°F supply water temp was around 3, which is pretty good. But if you could run it down to 120°F, you'd have a COP more like 4.5. By doing some minor plumbing changes, they're able to run it consistently at 120°E That's a substantial improvement, and we wouldn't have noticed it without the data.

FG: We wanted to know where it's warmest in multifamily buildings. Because of the stack effect, we expected the top floor apartment to be warmest. But .from what we can tell so far, a majority of the time the bottom floor apartment is warmest. Similarly, on the domestic hot-water (DHW) side, we found 30% to 68% greater consumption than the ASHRAE industry standard numbers had indicated, which I think was a shock to a lot of people.

HE: Fred, you mentioned having 12 million data points. How much do you look at? Have you ever tried to do real-time analysis ?

FG: In one problem building, I sat in my office for 20 minutes and watched data come in from a hot water system every five or ten seconds. Using that real time analysis, I determined there was a problem with the hot water coil. We found it, and it needed to be replaced. That was different from most studies, which look at data anywhere from a few weeks to months or years after it's recorded.

I'd say I probably look at haft the data in any given project. Since you have the ability to monitor more stuff, I say, let's take it. You never know what you'll find a use for. In our big DHW study, the first time around we monitored but didn't look at the heating side of it at all. Now, in a second or third project we're using five-year-old data, looking at temperatures, fuel flows, and burner run times. From a research standpoint, the data is just as valuable today as it was five years ago.

DP: Our largest project thoroughly analyzed a little over 9 million data points. Sometimes we'll throw all the data up on a graph initially, to make sure the values are in the proper range and the performance is as anticipated. Once things are running, we have an automated process for screening the data. A simple program looks at the data and compares ranges. For example, if a compressor is running, a fan should also be running. The program verifies that those run together and simple things like that. It allows us to evaluate the data as it comes in and catch problems and correct them before we get too far.

We usually check each site when it's instrumented. Since we're using Campbell data loggers, we can call it up via PC 208 software and look at it real-time to make sure it's OK. If there's a problem with something, or an instrument appears flaky, you just call it up on the phone to see what's going on. Often we'll have somebody at site looking at the data while I'm looking at it remotely. There can be somebody down in Miami working on a house. They'll call us and say, "See ffI can fix this thing and you watch the data."

FG: You need to make sure you're collecting the right data to answer your questions. But if you have the capability to take additional data points or greater detail, take it.
 

This single-channel Hobo-Temp temperature logger provides programmable digital recording capability and is simple to use. This self-contained unit sells for under $ 100, and is the property of the Vital Signs Project. Vital Signs maintains a"lend-ing library" of monitoring equipment that can be checked out by the staff and stu-dents of architectural programs at the University of California, Berkeley.

DS: Especially with remote monitoring projects. It might cost you 15 minutes to add a data point when you're at the site. Once the equipment's installed, adding a data point could cost days. We write a detailed monitoring plan that outlines what objectives we're trying to achieve and how to achieve the objectives. We boil that down to a data point list.

DP: With each project, we have a list of measurements we need. We also leave an analog and a pulse count channel open in case something comes up that we didn't expect. In terms of data overkill, I've been in projects where more data was taken than was necessary. But when reviewing that data, you can always dump the excess and keep what's useful. You need a data management system that lets you use the data you have. On most projects, analysis is very time-intensive. It may take two weeks or longer to satisfactorily analyze, interpret, and write up the information.

We have about 25 buildings; we're collecting 20 to 30 points of 15-minute data on each. Several of those have been going for three or four years. The data is reviewed by software--that is, when we collect the data nightly, it's brought in over telephone lines and sifted by the VAX mainframe we use here to collect the data. It checks for ranges and some other logical checks. We do things the hard way, in that we have a project engineer in charge of each project who reviews the data. They come in each morning and there's a plot waiting for them. It only takes a few minutes, and catches a lot of things that have gone wrong.

What's often forgotten is how much time is involved with doing a good job on the analysis. You have to manage large amounts of data and navigate through it. You need good software and you need to know how to use it.

FG: I, or a project engineer, usually download a building's data every three days and look at it the next morning. It takes a couple of hours the very first time, but you get to the point where you can sift through a few hundred kilobytes of data, which is God knows how many data points, in five minutes, or even two or three minutes if it is clean.

DS: It's easy to get sucked in by the data. I couldn't tell you how many hours I have spent pursuing some little
deviation in the data down a path, looking at it in different ways. It's a lot of fun, and you learn an awful lot, but it can really absorb your time.

HE: What fraction of time do you spend installing the equipment, debugging, and collecting and analyzing data ?

DP: Depending on its complexity, we have people who can install a site in between a halfa day and a full day. That would be a 20- to 30-channel project. Reviewing the data takes only a few minutes a day for the project engineer. Analysis is a long process and as Dave mentioned, you can get sucked into working on some specific problem. You sometimes go down roads that aren't fruitful, trying to troubleshoot something that you don't understand.

FG: We've cut down on the first steps, installing and debugging, by using, for the most part, existing building systems. The debugging part is probably the largest part. It's much larger than the installation. Collecting and analyzing takes probably 80% to 90% of the time. Sometimes that can go on for months.

DS: A day and a half to two days for a 30- to 40-data-point monitoring project would be typical. That includes calibrating the sensors, checking out data logger programming and modem operation, and everything else. There's always something that comes up that costs more time than you expect. Data collection and error checking, per building, takes maybe five minutes per day. Analysis time depends on what you're doing with the data. For the geothermal heat pump project, a report of COPs, energy usage, and that sort of thing is easy. Developing graphs from data is a snap. Trying to project energy savings from the data is most time-consuming.

HE: Do you have any data logger stories from Hell ?

DS: Running monitoring system wires through an attic in Palm Springs in July is the closest I've ever been to Hell.

DP: For a lot of our experiments on houses, we try to get answers on how something affects space cooling. The way we do that now, until we become brighter and find a better way, is we make a change right in the middle of summer and compare the before and after states. That involves doing a changeover during the second week in July, when it's pretty miserable. The summer before last, a radiant barrier was installed at the end of June. The installers planned to show up at 4 am, but they got to the house at 5 or 6. The way things get delayed, they were still up there at noon. One installer passed out and had to be taken to the doctor. The attic air temperatures were being monitored in real time, so we know that when he passed out, itwas almost 130°E It's amazing he was able to be up there at all. If your definition of Hell is thermal, that would be pretty good.

I also think of our experience using a donated data logger that someone wanted us to use in a project. The thing didn't work very well, and we lost most of the data as we went along.
 

This data logger (small box to immediate right of computer) is being set up to measure input power to an air conditioning unit. The computer and voltmeter (far right) are used for set-up and testing, but are not left at the site during monitoring.

FG: Always look a gift horse in the mouth. I worked with a client using 30 donated data loggers. These units take 4 data points. They were some of the most antiquated equipment I've ever used. They constantly went out and could not be reprogrammed on site. Many of the newer products I'm familiar with can be programmed for whether you want analog or digital input and all the other parameters. These have to go back to the factory to be reconfigured. You can try to hook three of them together to take 12 channels, but you have to use a phone switch. The manufacturer sold us a local area network to connect three data loggers in one building, but its outmoded design rendered it useless with currently available modems. The manufacturer finally suggested a third-party device, which they didn't even know how to hook up. They had us call an old client who helped us figure this out. It's nice to get a free data logger. It saved us maybe $30,000 or $40,000 in equipment. But it probably cost us three or four times that in man-hours, repairs, and hardware upgrades. There are also gaps in the data caused by the data from malfunctioning units, which makes the analysis hours go up considerably.

HE: If an HVA C contractor comes to you and says, "I'd like a bit of data-logging equipment in order to do troubleshooting," what would you recommend? That includes both hardware and software.

DS: We actually did a project with a contractor, and he wanted to take some of the equipment back with him. So he bought one of the data loggers that was used in the project. I gather he's been using it for troubleshooting. This was a Data Electronics unit. What made it attractive to us was that it allowed us to do a lot of programming. And it had a lot of "depth," so if you want to do logical statements in the programming you can do that. The key for a contractor would be the ability to measure temperatures easily, and perhaps extensive measuring equipment, like thermocou-pies, that connect directly to the logger.

HE: You don't recommend HOBOs or stand-alone equipment ?

DS: The HOBOs are great until you're measuring more than a couple of data points. By the time you buy the multiple HOBOs you need to get the job done, you could have bought a larger, more powerful system.

FG: Has anyone used the ACR Smart Readers? You can take up to eight temperatures with one of those, which seems nice. You don't need extra wires or phone lines. You install them and let them take data. After a time, you pull them and download the data to your PC.

DS: They might be the exception. We connected them to power monitors and used them for monitoring swimming pool pumps. However, the phase in the scanning varied enough from the phase on the power monitor output that for the first two weeks we got garbage. So that's something I'd be careful of if you're measuring power with those things. I'd recommend testing out equipment combinations before installing them in the field.

DP: The ACR loggers are good, I think, but they are expensive. But it depends on what you're trying to do. If you have very few channels, and it's always going to be the same channels you want to log, they're fine. These days, I'm leaning toward the Pacific Science and Technology loggers. They're cheap, easy to deploy once you have them set up, and they've been pretty reliable. They don't have anything to measure temperature, although we've been using one from a different company that's quite a bit less expensive than the ACR units. That's the Avatel T-101 high-temperature logger that sells for $220. You can collect about 120 days of data before you need to go retrieve it. We're using them in a project where we're trying to do temperature measurements inside homes inexpensively. Something that you can deploy quickly is always attractive.

Most of the time, what people from the A/C industry are interested in is "Can you recommend a portable unit I can use to obtain temperature and enthalpy measurements before and after the coil?" and so forth. There is a variety of equipment that will allow you to do that.

HE: How about in kWh meters?

JB: I like the Pacific Science and Technology Line Logger. It reads volts and watts, as well as kWh, and seems fairly straightforward. However, it retains the last reading indefinitely, so if customers don't reset it, they're adding new readings onto old. It's just a single data point kWh portable meter that plugs into the wall. You can hold it in the palm of your hand.

DS: We're impressed with Synergis-tics' programmability. They're pretty idiotproof and have built-in modems. It's easy to reconfigure the analog channels for different functions just by throwing a switch.

DP: I would agree with that--they're very easy to set up, easy to configure, and are nearly idiotproof. But having used both those and the Campbells, we prefer the Campbells. For certain measurements, especially very low-voltage devices like a pyranometer, you have a problem with the Synergistics. Also, with the Campbell, you can program it to give output in engineering units like degrees Fahrenheit and watts, rather than volts and pulses.

DS: I think they're best when you're monitoring lots of loads. 
 

The OAS Heat Computer can both control and monitor a building's existing boiler control system, saving time and money on installation of data loggers.

HE: What loggers haven't worked so well? 

DP: Unreliable sensors are a big problem. For example, flowmeters fail. Any insertion or paddlewheel type must be calibrated in situ to get anything reliable out of it. We use displacement meters, although they usually fail within two or three years. So if you've got 20 or 30 of them out there, you have one fail every month. The reliability becomes more important as a meter is an air trip or a full day's journey away. It may only take a minute to get in and replace the meter, but travel time adds up fast.

Hygrometers by most manufacturers are worthless. Only a few are worth anything. Chilled mirror types are most accurate, but require constant maintenance, so they're not really appropriate for field deployment. We've had most luck with Vaisala for measuring relative humidity. Other companies' meters saturate and don't come back. Also, lightning strikes cause most failures. The last logger we lost, we saw solar insolation go down to about 100 watts per square meter because a big thundercloud came, and poof---no data logger.

DS: We've had pretty good luck with flowmeters until our current geothermal heat pump project. The ones we have trouble with are the ones in the ground loops. The loops are charged with methanol, and require a big piece of charging equipment to replace the water once you've broken the line. We've had three fail within two to five days after installation. The broken ones don't seem to fail for any particular reason. There is a direct relationship between the liklihood of a device failing and how hard it is to access.

FG: We've had exactly the opposite experience with ISTA flowmeters. If you pay a little more, you can get ones with less pressure drop. We've found these meters to be quite reliable in hot water, cold water, and fuel. One has been running in heavy #6 fuel oil for six years now and has needed cleaning just once, after four years. Water meters have been extremely reliable.

DS: Thermocouples are risky with long wire runs. We see a lot of grounding problems.

DP: Differential thermo-couples, rather than single-wire runs should solve that problem.

DS: Using heavy gauge type-T thermocouples also helps.

FG: Even using shielded cable and type-T thermocouples, we've had trouble running them to the top floor of a six-story building. Another chronic problem is that some units provide output that's encrypted in a way that only their proprietary software can understand. Anyone shopping for data loggers should be sure the units they buy provide output in unencrypted ASCII format, so it can be used in any computer program.

HE: Is there any other lesson for people who might use data loggers ?

DP: With real estate, it's "location, location." With data loggers, it's "experience, experience." Always talk to someone who has used them before you do your own project.