Wednesday, November 25, 2009

Propane Use and the Power Conversion Number

In the prior co-generation posts an important number I've used to calculate the propane to power factor via the Ecopower. That number is 5.82 KWH/gallon of propane (plus about twice that amount in heat produced which is also used). That number comes from the Ecopower spec-sheet that has the rate of 4.7KW with a fuel rate of 3.42 lbs/hr, and using 4.24 lbs/gal, the weight of propane, you get the 5.82 number above.

I've been putting off the analysis of trying to check the power conversion number, mainly because it meant aggregating my propane bills and importing the numbers into a spreadsheet. Well I did that finally and this post goes through the estimation.

Our propane tank has a 500 gallon capacity, and about once a month the propane truck shows up and fills the tank to some level near the top. Exactly where it gets filled to, I don't know, but I know it isn't to full capacity. They always leave a little room for vapor expansion and such, so there is potentially some jitter there on a month-to-month basis. I will make the assumption that the tank is filled to the same level each time, and the amount of power generated from fill-to-fill is related to the amount put into the tank.

Another complicating factor is that we also use propane for cooking, and the amount of cooking isn't necessarily a constant, so if we just take the amount of propane used and divide that into the amount of KWH generated, the result will be an under-estimate of the conversion efficiency, and there will be a fill-to-fill variation depending upon how much stove top usage there is. The only way to get a more accurate number would be to put a flow meter on the co-generator fuel line, and I may request that, but the soonest that would be done is January, 2010.

The useful data is comprised of 16 periods starting with a fill on July 22, 2008 to the latest fill on November 9, 2009. The following figure shows the daily propane usage for the periods where each box represents one fill.

The red line in the figure is an estimate of the baseline propane use, and at just under 4 gallons per day. I'm going to call it 3.9 gallons/day baseline which is where the line is drawn. That sure is a lot of fuel just to heat hot water. When we were using heating oil, we used about 1.9 gallons per day for hot water use which is still quite a lot, and it continues to bother me.

At any rate, how do the numbers compare? For starters, we have to compare with similar heat content. A gallon of heating oil, if burned, releases 139,000 BTUs of heat while a gallon of propane releases 91,333 BTUs, so heating oil has 1.52 times more energy than propane, gallon for gallon, which means that if magically we could run the Ecopower on heating oil (or diesel for that matter), we would be using 2.56 gallons of heating oil--still above the 1.9 gal/day old use.

Well, we are close to our heating oil use, but we have one more adjustment before a proper comparison. Because we are co-generating electricity, we using 27% more than we would normally use in heat production, and if we divide 2.56 by 1.27, we get 2.02 gallons per day of heating oil boiler equivalent--close to our old baseline rate, so that's good on a consistency basis. It is a little bit high, though, and that could in part be explained by the cooking use. Let's keep that in mind as we continue on our goal, the propane-to-power conversion number.

The other great thing about the Ecopower is that it keeps track of how much power has been generated, and I've been logging that information too. The next figure shows the KWH generated per gallon of propane for the 16 periods. The blue points are the raw numbers without any adjustment. The x-axis has the number of days between fills, partly as a way to separate the estimates, and partly to look for a bias that would show the cooking use. The red line is the advertised value from Marathon Engine, 5.82 KWH/gal.

Looking at the data scatter, one can argue that there is an average daily use, and it works out to about 0.28-0.29 Gal/day cooking use. This number is determined iteratively. Frankly, I haven't thought much about cooking use, so I don't know how to interpret that value, but that value would give an agreement with the old baseline heating oil use of about 1.9 gal/day (in fact, it overcorrects!). The blue line shows the least-squares fit to the data, and it intercepts out at 5.62 KWH/gal which is the number that should be compared to the advertised value. The error estimate on that is 0.1 KWH/gal (from a bootstrap technique for any geeks reading this).

One has to be careful over-interpreting these data. The two points out near 80 days between fills were early on. After those points were taken, a leak in the system was found. Was the leak present during those points, and later use has less of an effect? Also, the laggards at 20 days and near or below 5 KWH/gal can be selected out, because my in-laws were in town, and my mother-in-law loves to cook. Could she have been using 4 times as much propane per day we normally use in cooking? Seems easy to believe, and it can't be ruled out at the moment. If you discard those two laggards, you get to the 5.82 value, and it would raise the daily cooking use to about 0.4 gal/day. All sorts of data adjustments are possible. I'll leave the values as I found them for now which is the appropriate thing to do.

This post has tried to test the Ecopower propane-to-power conversion number advertised by Marathon Engine. The value I get using the number of gallons reported to me by the propane delivery company and the number of KWH reported by the Ecopower is 5.62 KWH/gal plus or minus 0.1 KWH/gal, and I estimate we are using 0.29 gal/day of propane in stove top cooking (our oven is electric). The measured propane to KWH number is lower than advertised by about 3%, but it is within the statistical error range and probably within the possible systematic errors, the bulk of which would bias our estimate to be too low.

Note: The daily cooking rate number was originally calculated incorrectly, and I have corrected that from an earlier version of this post and adjusted the text appropriately.


Saturday, November 21, 2009

Co-generation Operation Cost vs. Fuel Cost

In an extension of previous posts, I've looked at the Ecopower operating costs versus fuel costs for propane and natural gas. The above figure shows the cost of power production for these two fuels. There are three different lines drawn. The blue lines show the cost of production assuming that no "waste" heat is used, and clearly the costs are rather high.

The red lines (and next lines down from the tops) show the generation cost in co-generation mode. The main assumption is that the extra fuel requirements are 30% over a traditional boiler or furnace.

The lower purple lines in both graphs in the figure is the equivalent market generation cost in the case of a $0.05/KWH delivery cost. Currently here in CT, the delivery cost is $0.0545/KWH. The purple lines show what the market competition is (coal!). What is notable is that with a propane price of less than $1.5/gal or a natural gas price of less than $15/mcf (retail price), it makes sense always to generate the power locally in co-generation mode if there is a net metering arrangement in place, because it costs more to deliver the power than it does to co-generate the power locally.

Current costs this year in CT is about $2/gal for propane and $20/mcf for natural gas. This is the retail price, and for natural gas it is rather high as wholesale natural gas for delivery next month is currently under $5/mcf. There is always a retail mark-up due to the extra delivery costs and, in the case of natural gas which is regulated, there is probably an additional cost from prior hedging with gas procured for delivery earlier in the year or last year when prices were significantly higher. The retail power cost including generation and delivery is in the range of $0.15/KWH to $0.18/KWH depending upon the retail provider (we have choices here in CT).

The conclusion is that co-generation wins hands down, and the primary reasons are (1) close to 70% of the fuel energy is lost as waste heat up the smoke stack and in transmission, and (2) there is a delivery cost for distributing the power, and these extra costs are larger than the extra retail fuel cost over wholesale fuel cost.

I would love someone to check these numbers. The methodology is described in prior posts on the co-generation costs.

Labels: ,

Thursday, November 12, 2009

Time of Use Power Pricing

The current power rate structures that exist provide very little options for power users. One of the reasons is that historically power meters have to be read on-site and are usually done so only once a month. The only real option that users have had are either a single rate structure where power costs the same regardless of when you use it (what most everybody does), or a time of use structure where you pay less for off-peak usage and more for on-peak usage. Here in CT, peak time is between Noon and 8:00 PM local time weekdays.

Here in Southern CT, the peak/off-peak option doesn't get one much if anything. We are somewhat deregulated here, so there are a few companies vying for retail power business. The main supplier is Connecticut Light and Power, but there are other ones now, and most of them offer rates that beat the peak/off-peak CL&P rate, so if one is interested in saving something, it seems best to just switch providers and take their single rate structure--which is what all of them only seem to offer.

The one new difference now is that CL&P offers something called Variable Peak Pricing or VPP. The VPP option involves a fixed off-peak rate and a peak rate that varies daily. The VPP option has been available for about a year. I decided to look into VPP just to see if it offered any advantage, and here is the result.

The first question was if the pricing could be reproduced. The off-peak price is just the same as the standard Time-of-use option, and it is presently about 11.5 cents/kwh which is more expensive than some of the competitors' single rate cost. For standard TOU service, the peak price is about 14.5 cents generation rate. To get the billing price, you have to add $.0545/KWH, the transmission and taxes portion of the price.

The VPP history is posted on the CL&P website, and it is shown below. Included is my attempt to model the price, and I can get the variation, but there is also an offset of between 2 and 5 cents that seems to reset every quarter or so. Exactly how this offset is determined, I don't know, and I have a call into CLP and hopefully will get an explanation. At any rate, to within an offset, I can reproduce how they get their price just by taking the Day-ahead New England price and averaging over the peak demand hours (Noon-8:00 PM).

The first item of note is that the peak price is in fact lower than the set off-peak price and has been for about a year now. At present, then, it makes sense to shift load to peak times as opposed to off-peak times! That is a characteristic of how CLP currently structures their program as well as lower natural gas prices this year as compared to past years and frankly doesn't make any sense. I'm guessing it is a result of CLP hedging their off-peak price when power prices were higher.

Given the historical day-ahead pricing data from the New England ISO, we can now estimate how the price would have varied back to 2003 and that is in the above figure in blue. I've assumed an average additional $.035/KWH as the average offset fixed over the period. In the more distant past, it is possible that that figure is too high, but it is probably reasonable going forward. The black line is the actual single-rate price from our electric power bill. Only the generation rate is shown, and transmission, taxes, etc would have to be added to get the true cost. The off-peak price would probably be at a discount to the single-rate price, probably on the order of a penny or so.

The question still remains as to whether or not to shift to VPP. It seems that it makes sense in a market where the power prices are falling, and to stay with fixed pricing when power prices are rising, which is the standard result of hedging--it works in rising markets, but you pay for it in falling markets.

For most people, I think it would make sense just to shift to a different provider for now to save money. However, with some load shifting capability (and with generation capability), VPP may even make sense. One needs historical off-peak prices to answer that question, and I have that call into CL&P for that too.

I was also hoping VPP offered the ability to do daily-net metering. With generation capacity (both solar and co-gen), there are time where I could crank up the amount going back to the grid. Unfortunately, net metering occurs only over the course of the year, and so there is no advantage to generating more than would be used on a daily basis. For instance, in early January, 2004, the price was over $.4/KWH. This cost is large enough that I could run the Ecopower at full throttle and still make money if CL&P was in fact paying $0.4/KWH for power sold back. However, under the VPP program, the most economical thing I could do would be to run enough to offset any purchases made that day so that the net grid power purchase was in fact zero.

The question of whether or not it is advantageous to switch to VPP remains and to better answer it, I need a daily load model for our house. With some data I've collected, it is possible to generate that, but it is going to take more work.