The heating value of a volumetric measurement of gas is the single-most defining property in determining its energy value—and its economic value. Consequently, uncertainty and inaccuracy in determining the heating value of a volume of gas leads to the same problem in financial accounting of that volume.

To assess the impact of using inaccurate or non-representative data in gas accounting processes, Telvent GIT SA solicited its customer base for real-world data. One company, "Company 2," provided a year's worth of actual gas volume, heating value and energy value data. Telvent scrutinized this representative data set and produced a case study based on its findings.

Telvent is a global information-technology solutions and services provider with North American headquarters in Rockville, Maryland. It serves every energy company in the Fortune 100, and its information management solutions help manage and measure hydrocarbon movements in the majority of North American and Latin American pipelines.

Case study data shows the time series of gas flow heating value (red series). Heating value is significantly higher than average for about 200 hours of the 5,200 hours examined, according to the figure inset. Note the diurnal pattern in the volume values (blue series). For some measurement times, volume of flow is essentially zero.

To begin the study, Telvent profiled the actual variability in the heating value of gas flow. It shifted actual measured heating values to subsequent measurement periods, thus simulating the effect of delayed application of the gas properties to the measurements. Also, it simulated the effects of using non-real-time or less-accurate heating values, such as composite-sample data or spot-sample data, in calculating the gas energy value by aggregating actual values to synthesize composite sampling and choosing a single value to be applied across a range of records to simulate spot sampling.

Comparing the actual correlation of volumes and heating value data to the simulations, Telvent identified errors that can occur in the real world when inaccurate heating values are delivered to the flow computer or business office to determine the gas volume's energy and financial value.

The case study attempts to answer questions such as: Should an enterprise adopt a different, more accurate means of determining gas quality and heating value, to avoid inaccurate energy values? Should it consider a different means of integrating flow data?

Types of analyses

For the purpose of this article, the term energy is considered to be the product of volume and heating value. Industry organizations such as the International Standards Organization (ISO), Gas Producers Association (GPA), the American Petroleum Institute (API) and the American Gas Association (AGA) have technical standards defining the methods for determining heating value and energy content of flowing gas. These standards cite gas composition analysis as a key determinant of heating value. Three primary gas sampling methodologies include online, composite-sampling and spot-sample analyses.

The heating values measured over the case study year demonstrated bimodal distribution, with the gas flow reflecting a high occurrence of two distinct heating values, compared to other values. This phenomenon negates the use of statistical analysis techniques assuming normal distribution.

The online analysis is the most certain way to determine the heating value, and calculated energy value, of flowing gas. The online gas chromatograph (GC) automatically samples gas continuously during flow, measuring composition at frequent intervals in a continual process. Its key attributes include accuracy and high resolution, but the approach is expensive, requiring a complex field installation and ongoing maintenance.

The composite-sampling analysis method collects samples at intervals with an analysis of the composite of these samples completed by off-site gas chromatography. While not as comprehensive as continuous online analysis, composite analysis is representative of all of the flowing gas. Its key attributes are that it is less expensive than online analysis and that it might yield a more detailed analysis as a result of more sophisticated chromatography in the laboratory. Also, it requires a rigorous sampling regimen, including some installation of field equipment.

The spot-sample analysis method collects discrete gas samples periodically for manual analysis at an off-site laboratory, as with composite sampling. Because considerable time can elapse between these discrete samples, this methodology might not be representative of gas flowing at the time of reporting. However, it is the least-costly composition monitoring, could yield a more detailed analysis resulting from more sophisticated chromatography in the laboratory and typically produces the most uncertain gas composition.

Of the three methods, only online, or 'live,' analysis can deliver real-time heating values to the flow computer and to a centralized system when polled through remote telemetry.

Shifting the actual heating value (red series) shown in Figure 1, to volumes measured 178 hours after real-time gas analysis, results in energy value calculation error (green series).

Variability of heating value

The first step in determining the best analysis is to determine the statistical properties of heating value in actual measurements. A number of methods can be used to characterize the data.

Heating value can vary on its own. Figures 1 and 2 illustrate the case study data during the year. These data show an average heating value of 1,026 Btu per cubic foot, with a maximum value of 1,045.8, a minimum value of 1,014.5 and a bimodal distribution.

Analysis of composite or discrete gas samples introduces, by nature, some amount of delay between actual flow of the gas sampled and the availability of the analysis results of that flow. So, what is the accuracy effect, or business impact, of using a delayed heating value for energy determination?

Telvent took the heating value profile illustrated in Figure 1 and simulated how a delay in heating value availability would impact the accuracy of the corresponding calculated energy value. Figure 3 shows the error in energy calculation when the heating value profile is applied to volumes measured 178 hours later, a delay of about one week. The error shown is the percentage difference between the energy calculated using a delayed heating value and the actual energy value calculated in the field.

This shows the range of percentage error, when energy value is calculated with delayed gas quality data. The error reached bounds of +/- 2% as delay reaches 50 hours, then it does not range much further, positively or negatively. This indicates that the delay in using the correct heating value has an effect quickly, but then does not get appreciably worse. If a certain delay is inevitable and acceptable, then using a default heating value instead of downloading an out-of-date value might be the better course.

Figure 4 shows the extremes and average over the year, of the energy value calculation error as the delay in application of measured heating value data increases. Note, the error increases rapidly at first, and then less quickly.

Energy errors due to sampling

Telvent simulated a composite-sampling scenario with the original case study data by calculating the flow-weighted average of actual hourly heating values over certain time periods. In this manner, the company simulated composite samples reflecting consecutive 12-hour periods over the course of the year's worth of data. It repeated the simulation for consecutive 24-hour periods over the year, and likewise, for 72-, 168- and 770-hour periods over the year.

Then, it applied the weighted average heating value for each time period to the actual gas volumes during that period to see how accurately the period-average heating value represented the energy value of the gas flow during that period.

Spot sampling was simulated by selecting one heating value record every 10 hours over the course of the year. Telvent repeated the simulation using one heating value record every 25 hours over the course of the year, and again at every 75, 165 and 770 hours over the course of the year. The selected spot heating value was applied to the actual gas volumes during the time interval it represented.

Figure 5 shows the error in calculated energy value based on analysis of composite samples and spot samples, compared to the energy value calculated from the original real-time flow analysis.

Generally, the error increases as the composite-sample period increases—but not in a linear or predictive manner. This simulation, using actual hourly gas quality values, suggests that the longer the amount of time a composite sampling device is active on a pipeline, the more error will be seen in the energy value calculated from composition analysis of the composite gas sample obtained.

Energy-value calculation based on spot-sampling analysis results in larger errors than that based on composite-sample analysis. This is to be expected, assuming the composite sample better represents the gas flow over the time interval of interest. Figure 6 illustrates that it is also important to apply the composite sample value to the flow from the time the sample collection started. If the composite heating value is applied only to the flow occurring after the laboratory results are available for upload to field measurement devices, then the error is similar to that of a spot sample. For this reason, the final energy calculation must be done retroactively and off-site, in a back-office measurement system.

This figure shows the average error in energy value calculation based on heating values determined by analysis of a composite sample (green series) and of spot samples taken at intervals (blue series).

Break-even

In this case study, the measurable quality of the gas supply varies as the supply flows through the online gas chromatograph or sampling device. Consequently, any sampling or calculation method substituting for online, real-time analysis will result in gas quality reports that will err, to some degree, from actual values at any hour of the supply. As the composite sample collection period increases, or as the spot sampling interval increases, the error of the energy value calculated from sample analysis increases.

Depending on the heating value variability in the gas supply and the sampling period or interval chosen, there might actually be a zero apparent error over the course of a certain delay or sampling period. That is, the sampling period or delay might "tune" with gas quality variation. Such tuning is not a desirable effect and should be considered before implementing a gas quality management program.

This analysis of a composite sample collected over a 168-hour period shows a calculation of an energy value that is less in error (green series) than that based on spot samples collected every 168 hours (blue series).

Composite or discrete sampling might adequately represent the actual supply at any time if the sampling regimen is chosen to best reflect the potential variability of the supply and the accounting period, be it daily, weekly or monthly. In this case, the supply company could minimize the costs related to more frequent sample collection and gas quality analysis, and see minimal energy value calculation errors sent to the financial accounting process.

According to this case study, heating value of the gas flow does not vary uniformly or predictably, and the costs associated with establishing and maintaining online gas composition and automated, end-to-end accounting might be returned by minimized energy value errors in the accounting process.

Bill Morrow is a product manager at Telvent, responsible for the company's software products for the energy industry, including applications that aid operational automation, measurement and forecasting business processes.