Many gas piping system simulations involve the use of load data based on customer billing records. A common practice is to extract pertinent information from the mechanized customer billing history record and from that information, to compute a load (demand on the piping system) from the billing information. Such customer loads are then (possibly grouped) assigned to nodes on the piping system in order to get a spatial distribution of total system load within the piping system. This paper is primarily concerned with pointing out sources of errors in this process. Although the procedure of inferring demands at some instant or at some hour from billing records, is by no means perfect it is widely recognized that the procedure is the best available considering the costs of other methods. The purpose of this paper is not to denigrate the process, but to make the reader aware of the problems so that their existence can be taken into account in developing usable loads. Most gas system operators have available on an hourly basis, total system load. In some cases, good estimates can be made of what total system load will be under varying weather conditions, time of day, and season of the year. If one were to divide this total system demand by the number of customers, then we would arrive at an average load per customer. Knowing how many customers are served from each node in a distribution system the assignment of loads would be straightforward. Very few, if any distribution systems serve customers whose structure sizes don't vary significantly from the average nor whose personal usage rate of gas is so uniform that this simple average load is adequate for determining pipe sizing and piping design, especially when attempting to simulate actual operating conditions as opposed to designing a new system.

Fundamental Premises

The fundamental premises on which the use of customer billing data is based are recognized to be only imperfectly true. An appreciation of what these imperfections are might help in developing the proper logic to develop local node loads. Among these premises are:

  1. Past consumption patterns are predictors of future consumption patterns under the same weather or seasonal conditions.

  2. All consumption values from billing records are correct. This in turn infers a. All recorded meter readings represent the actual meter readings. b. Meters perfectly measure the quantities of gas which has passed through them (proof is 100.0). c. There is no unaccounted for gas in the system (no leakage, bypasses, unrecorded metered connections)

  3. The customers do not replace gas burning equipment with that of different efficiencies.

  4. Customers do not change thermostat settings or thermostat types

  5. Customers do not vary in their usage of hot water nor the number of clothes dryer loads per week that they run, nor change the hot water heater thermostat settings.

  6. The number of times doors are opened and closed (to the outside) do not vary.

This content is only available via PDF.
You can access this article if you purchase or spend a download.