Heating costs

In 2010, my heating costs exceeded my advance payments by about 25%. This motivated me to decompose the costs to see what drove the changes. Here is the result:

The numbers refer to Euros. Read von right to left: 2010 was a cold year (+102EUR), but gas consumption in this house was relatively low (–89EUR). Also, running costs and gas price were below previous year’s value (–49EUR). The main driver (+183EUR) for the increased costs was my share on the total costs. In 2009, I had to pay 1.7% of the total costs, which rose to 2.3% in 2010. This means my landlord assumes I used more energy than the other tennants in 2010. I doubt that, and one aspect to support my doubts is that the values of the two heat meters in my living room were nearly at the same value, while I used only one heater.

Thus I believe it was a measurement error. I am not going to complain about this, but will check next measurements very carefully.

Not a direct result of the decomposition, but also interesting is the (temperature adjusted) enery use per square meter, which was in 2010 by about 110 kWh/m2, which is nearly average.

Here’s the R code to produce the chart:

The data is an example data set of my pft package, the decomposition is performed with the ida package. These packages are not yet on CRAN (I consider them not mature enough). They are available in my personal repository and can be installed with

> install.packages(
+   c("ida", "pft"), 
+   repos=c(getOption("repos"),
+      "http://userpage.fu-berlin.de/~kweinert/R"),
+   dependencies=c("Depends", "Suggests")
+ )

Currently, only R 2.13 and R 2.14 binaries for windows are available.

We represent the heating costs as a product

C(t) = g(t) * w(t) * i(t) * s(t)

with

• g(t), the temperature adjusted gas consumption
• h(t), the relation between the heating degree days in year t and the long term average heating degree days
• i(t), the costs per cbm gas including running costs
• s(t), the share of total costs attributed to me

The dataset pft.heating contains these contributing factors, and is loaded with the data function. The decomposition is calculated with the ida function.

> data(pft.heating)
> decomp <- ida(decomp, effect="variable", 
+    from="Y2009", to="Y2010", 
+    method="lmdi1")

The resulting object decomp has a custom plot method. If the plotrix package is installed, it produces the waterfall chart above, otherwise a bar chart.

> plot(decomp, digits=1)