Modeling Energy Usage in Waterville


Peter Smithy ‘12

Noah Teachey ‘13


Environmental Studies Program

Colby College

Waterville, ME




The goal of this research was to calculate the energy load of buildings in Waterville and to determine the feasibility of building a 5 Megawatt power plant at one of two sites.  This would meet the heating requirements of the city through the piping of excess heat to the buildings.  We determined the heating load by multiplying square footage with an estimate of energy load per square foot per year and the paths of the pipes were based on preexisting sewer pipes.  Our preliminary analysis found that hypothetical power plant site 1 is more efficient in terms of piping distance to 433 commercial and residential zones than site 2.




The use of large power plants in the distribution of power across great distances is inefficient due to the loss of energy as low quality heat in the conversion of heat to electricity.  It has been demonstrated that in 40-55 MW power plants the second law efficiency varies from 18% - 23% (Kumar 2009).  In other words, potentially useful energy escapes in the form of waste heat, resulting in a relatively resource-intensive energy extraction process.

Rather than releasing excess heat as heat pollution, as is the practice of many large power plants, greater energy efficiency would result if it were possible to capture and utilize the excess heat energy.  Demand for such heat energy as a means of heating buildings incentivizes the construction of pipe networks from power plants to communities for which there is a demand for the heat.  This scenario loses feasibility the further a power plant is from the community that might benefit from the heated water vapor.  The purpose of this research is to calculate the energy load in Waterville, ME in order to determine the viability of heating the community with waste energy from a proposed 5 MW power facility.




After obtaining vector data including the square footage of buildings and the location of sewer pipes, we used GIS software to apply a figure for the yearly energy load (0.5 gal. of oil / sq. ft. yearly) to the square footage of buildings to estimate total yearly energy load.  In order to display the more energy-intensive areas of Waterville, we ran a Kernel density analysis in which the variable displayed was total energy load.  We selected two hypothetical power plant locations for comparison; one at the site of a railroad hub and another at the site of a former paper mill.  We used a zoning vector layer to sum the total energy load of buildings within each zone and calculated a centroid for each zone before measuring distances along sewer pipes from each hypothetical power plant location to the centroid of each residential and commercial zone.




In carrying out the analysis of distance along water mains from the two hypothetical power plants to the centroids of zones designated residential or commercial, we found that Plant Location 1 served 433 centroids by tracing 1,267.085 km of piping whereas Plant Location 2 accomplished the same task by tracing 1,355.560 km of piping.  However, Plant Location 2 used less piping than Plant Location 1 in serving the 36 zones designated Commercial A (primarily of downtown Waterville), with only 81.107 km of piping as opposed to Plant Location 1’s 92.813 km, as summarized below.


Table 1. Total pipeline lengths needed to serve the centroid of each zone for a given zoning classification. A new line is generated from the power plant to each zone. The number of zones served is equal for both plant locations and indicated.

Zoning Class


Downtown/ Commercial A

(36 zones)

Commercial B,C,D

(81 zones)

Residential A,B,C,D

(316 zones)


(433 zones)

Location 1





Location 2







Figure 1. Zoning map of Waterville coded by increasing energy load from light to dark red. The color range indicates a range of 50-350,000 gallons of oil used per year for heating. Brighter areas are localized hotspots of energy load, computed from individual building load.


Figure 2. A three-dimensional representation of energy load within Waterville, with rivers included, where higher elevations correspond to greater energy load.


Figure 3. A map of downtown Waterville, where buildings with higher energy loads are indicated by darker reds and hotspots of heating load are illuminated.





This finding demonstrates that the first hypothetical power plant location, near the railroad hub, is a more efficient location in terms of piping distance to all 433 residential and commercial zone centroids.  Piping distance relates to the quality of heat that can be extracted from the waste heat of second law-inefficient operations in a 5 MW power plant because heat is lost with every additional meter of piping through which it must travel to reach its destination.  However, if a greater degree of selectivity is used in building the infrastructure through which the excess heat is distributed and a lesser number of zones is served, then Plant Location 2 has the potential to be more efficient.  If, for example, a power plant project only produces enough heat energy for use in buildings in downtown Waterville then the second location is more efficient than the first.  There are a number of other variables that this research does not consider, such as the fixed costs of building the facility and complying with codes that might alter the optimal placement of the facility.  Additionally, the differentiated sizes of piping will play a part in further analyses of the data as trunk piping must be optimally placed to best serve the needs of the community with branch and connection piping feeding off of it.




       Hypothetical Plant Location 1 is the most efficient location in serving all 433 residential and commercial zones in the Waterville area along pre-existing water main pathways when compared with hypothetical Plant Location 2.

       Hypothetical Plant Location 2 has the potential to be more efficient in serving the Waterville area depending on which and how many zones are served.

       Consideration of differentiated pipes may change the results and will allow for a rough estimation of total costs of construction of the pipe network.

       This analysis represents a preliminary effort to display heating load and suggest the most efficient of two hypothetical power plant locations.  Actual pipeline locations will most likely diverge from preexisting water mains should this study be advanced further.




We would like to express gratitude to Prof. John Joseph of the Mid Maine Sustainability Coalition for suggesting the need to explore this topic and for providing several figures, Paul Castonguay with the City of Waterville Assessor’s Office for Waterville building, road, and zoning data, and Jefferson Longfellow, P.E. with the Kennebec Water District for water main data.  Additionally, special thanks go to Professors Philip Nyhus and Manny Gimond for technical assistance.