Measurement Uncertainty and Compliance Evaluation Applied to Natural Gas Moisture

Abstract

The reliability of natural gas moisture measurements is crucial in preventing corrosion in pipelines and equipment, ensuring burning efficiency and mitigating operational risks, and is often mandated by standards and regulations. An important quality parameter that aids in conformity assessment, particularly in risk assessment, is measurement uncertainty. The assessment of measurement uncertainty, based on the Law of Uncertainty Propagation, depends on the mathematical model used to calculate this physicochemical property. This study aimed to compare different algorithms for calculating moisture content in natural gas, estimate and validate the measurement uncertainty based on the algorithms implemented in the Portable Moisture Analyzer (PM880) equipment, a portable hygrometer manufactured by Panametrics in Wilmington, NC, USA, and evaluate compliance with Brazilian legislation using guard bands as a decision rule. The moisture results in natural gas varied by a maximum of 1% among the three approaches presented. Furthermore, based on the dew point and pressure results, the expanded uncertainties of moisture were about 20%, which did not compromise the risk assessment for the consumer, as the moisture results were well below the specification value. Consequently, the upper tolerance limit of 58.4 ppmv H₂O was established.

1. Introduction

The water vapor in natural gas, commonly called moisture, must be monitored with high reliability to ensure the gas’s safe processing and transportation. Moisture control is critical in processing plants, as it can significantly impact the quality and efficiency of transportation and lead to corrosion and other types of damage to equipment and pipelines.

The water content in natural gas is the primary factor influencing internal corrosion. Even at low levels, variations in pressure and temperature can cause water to condense, resulting in corrosion, hydrate formation (a semi-solid combination of hydrocarbons and water), or ice formation. To mitigate these problems, natural gas companies invest in dehydration units. Based on international standards such as ASTM D5454 [1], ASTM D1142 [2] and ISO 18453 [3], the design and cost of these installations are influenced by the need for accurate knowledge of the water content at the dew point and the contractually required water content, as natural gas pipelines have specifications that limit the maximum permissible water vapor concentration.

For instance, a standard pipeline specification in the Petrobras offshore production system is a maximum moisture content of 60 ppmv. The water dew point in gas refers to the temperature, at a specified pressure, at which liquid water will begin to condense from the water vapor present. Consequently, water concentration levels can be measured based on natural gas systems’ dew point and pressure.

As stated in ISO 6327, “the partial pressure of water vapor in gas samples is the saturated vapor pressure corresponding to the observed dew point, provided that the gas in the hygrometer is at the same pressure as the gas at the time of sampling” [4]. Therefore, to conduct moisture measurements in gas, it is essential to establish a correlation between the water dew point and the water content.

Shkarovskij and Volikov improved gas quality management, proposing a procedural framework for accurately determining the moisture of natural gas using dew point temperature indicators found in fuel certificates [5].

Studies conducted on hygrometers, generally, have shown to be an area of interest by researchers. One classic study evaluated the uncertainty in measuring the dew point of natural gas using the traditional cooling mirror condensation hygrometers [6]; another one compared various hygrometers used for measuring moisture in gases that are utilized with varying frequency in the natural gas industry [7]; the last study centers around hygrometers being used for measuring water vapor (moisture) in natural gas, focusing on their response to co-exposure to ethylene glycol [8].

Gallegos et al. [9] compared eleven hygrometers at natural gas facilities in Spain utilizing various measurement techniques (including chilled-mirror, electrolytic sensors, spectroscopic analyzers, and polymeric and metal oxide humidity sensors).

Chinese researchers have developed an optimized Gaussian process model that effectively evaluates parameters to predict the water dew point in natural gas dehydration units, enhancing the efficiency and accuracy of the dehydration process [10].

A conference paper illustrated that an online dew point analyzer operates within the lowest error range after a six-month recalibration period compared to other calibrated and reliable devices [11].

The National Physical Laboratory (United Kingdom) presented a method to isolate and adjust the influence of background gas composition on trace moisture measurements by employing two distinct frequencies in a trace moisture analyzer that utilizes a ball surface acoustic wave (SAW) sensor [12].

A recent study outlines approaches for increasing the quantity and ranges of reproducible values while enhancing the functionality of the state primary standard for relative humidity of gases, molar (volume) fraction of moisture, and dew/frost point temperature (DPT) as specified in GET 151-2014 [13].

Last year, Chinese researchers published an interesting study regarding an interdigital conductance sensor designed to enhance electric field distribution for the precise measurement of uneven liquid film thickness in inclined gas–liquid two-phase flows. Through experimental validation, the sensor showcased improved effectiveness over conventional sensors [14].

For an accurate assessment of natural gas moisture, it is crucial to employ standardized methodologies, ensure regular calibration of measuring instruments, and consider the uncertainty associated with each process stage.

This study evaluated the measurement uncertainty of natural gas moisture calculated from the dew point temperature and gauge pressure results using the algorithms implemented in the Portable Moisture Analyzer—PM880. An upper acceptance limit has been proposed to ensure that the moisture content in natural gas does not compromise the operational efficiency of gas pipelines. This limit is based on uncertainty data and employs the concept of guard bands to mitigate the risk of false conformity concerning moisture levels in natural gas.

2. Methodology

Several methodologies are used to determine the water vapor content in natural gas from the dew point temperature and pressure measurements. The first two parts of this section describe two methodologies, and the next part describes the methodology used here. The last sections are designated for calculating and using the measurement uncertainty for compliance assessment.

2.1. ASTM D1142 Test Method, Approach 1

The ASTM D1142-95 (2021) [2] test method includes two correlations, the first of which (herein referred to as ASTM D1142 test method, approach 1) is a variation of Equation (1) that expresses water content in terms of the weight of saturated water vapor, Equation (2):

$$y_w={\displaystyle \frac{P_w^{sat}}{P}}$$

(1)

where $y_w$ is the mole fraction of water in the vapor phase, $P$ is the total pressure, and $P_w^{sat}$ is pure water’s saturation (vapor) pressure at the dew point temperature T.

$$W=\left(w\times {10}^6\right){\displaystyle \frac{P_b\times T}{P\times T_b}}$$

(2)

where W is the water content (pounds water/million standard cubic feet natural gas) at reference conditions, $T_b$ (base gas measurement temperature, $t_b+400$), $P_b$ (base gas measurement pressure, psia), w is the weight of saturated water vapor (lb/ft³), P is the pressure at which the dew point was determined (psia), and T is the observed Rankine dew point temperature ($t+460$). The reciprocal of w, or the specific volume of saturated water vapor (ft³/lb), is listed as a function of temperature in Table 1 of ASTM D1142-95 (2021) for temperatures ranging from 0 °F to 100 °F. Although not explicit in temperature due to the temperature dependence of w, given the water content, the corresponding dew point temperature can be solved iteratively.

2.2. ASTM D1142 Test Method, Approach 2

Bukacek proposed a relatively simple modification of Raoult’s law approach, where the water content of the sweet gas is calculated using the ideal expression of Equation (1) supplemented by a deviation factor [15], Equation (3):

$$W=\left(760.4\right){\displaystyle \frac{P_w^{sat}}{P}}+0.016016\times B$$

(3)

where W is the water content (g/Nm³), $P_w^{sat}$ is the saturation vapor pressure of pure water (MPa), P is the total system pressure (MPa), and B is given by Equation (4):

$$logB={\displaystyle \frac{-1713.66}{T}}+6.69449$$

(4)

where T is the dew point temperature (K) and the saturation vapor pressure can be calculated by Equation (5) [16]:

$$ln\left({\displaystyle \frac{P_w^{sat}}{P_c}}\right)={\displaystyle \frac{T_c}{T}}\left(-7.85823\tau +1.83991{\tau }^{1.5}-11.7811{\tau }^3+22.6705{\tau }^{3.5}-1539393{\tau }^4+1.77516{\tau }^{7.5}\right)$$

(5)

where T is the temperature (K), $T_c$ is the critical water temperature (647.14 K), $P_c$ is the critical water pressure (22.064 MPa), and $\tau =1/T-T_c$. Thus, a simplified version of Equation (3) is Equation (6):

$$W={\displaystyle \frac{A}{P}}+B$$

(6)

Referred to here as Section 2.2, ASTM D1142 Method 2 is included in ASTM Test Method 1142-95 (2021). The coefficients A and B (referenced to $T_b$ = 520 R and $P_b=14.7$ psia) are listed as a function of temperature in Table 2 for dew point temperatures ranging from −40 °F to 440 °F. Although not explicitly stated in temperature due to the temperature dependence of A and B, the corresponding dewpoint temperature can be solved iteratively given the water content. Although conveniently simplistic, neither ASTM method considers the actual gas composition. Furthermore, the range of data available for the specific volume of saturated water vapor (ASTM D1142 Method 1) or for the coefficients A and B (ASTM D1142 Method 2) is somewhat limited.

2.3. Method Used by the Portable Moisture Analyzer—PM880

The Panametrics PM880, as shown in Figure 1, is a portable hygrometer manufactured by Panametrics in Wilmington, NC, USA. It employs an aluminum oxide sensor to assess the moisture content in liquids and gases. This measurement is based on the temperature at which condensation occurs under specific barometric pressure conditions.

The vapor pressure of water and ice, $Pv,$ depends on the dew point in °C, DP, and can be calculated from Equations (7) and (8):

$${Pv}_{water}={611.21}^{\left({\displaystyle \frac{17.502\times DP}{240.97+DP}}\right)} DP>0$$

(7)

$${Pv}_{ice}={611.15}^{\left({\displaystyle \frac{22.542\times DP}{273.48+DP}}\right)} DP\le 0$$

(8)

The 1958 IGT8 (Institute of Gas Technology research bulletin #8) calculation based on the dew point value is expressed by Equation (9):

$$IGT8={10}^{\left({\displaystyle \frac{-1711}{DP+273.15}}+6.687\right)}$$

(9)

The non-ideality of natural gas, B, depends on the dew point in °C, DP, and can be calculated by Equation (10):

$$B={\displaystyle \frac{IGT8\times P_v}{{Pv}_{water}}}$$

(10)

The total pressure converted to Pascals, ${Pressure}_{PA}$, is calculated by Equation (11):

$${Pressure}_{PA}={\displaystyle \frac{P{ressure}_{psig}+14.697}{14.697}}\times 101325$$

(11)

The ratio of water vapor pressure to total pressure, $X_{water}$, is calculated by Equation (12):

$$X_{water}={\displaystyle \frac{P_v}{{Pressure}_{PA}}}$$

(12)

Finally, W is the water content (pounds water/million standard cubic feet natural gas) at reference conditions $T_b$ (base gas measurement temperature, $t_b+400$), $P_b$ (base gas measurement pressure, psia), Equation (13), considering the conversion of lbs/MMSCF to a fixed value, ${\displaystyle \frac{{10}^6\times {30.48}^3}{\left(22414\times \frac{519.67}{491.67}\right)}}\times {\displaystyle \frac{18.0152}{453.59}}=47473.05721.$

$${W=X}_{água}\times \left[{\displaystyle \frac{{10}^6\times {30.48}^3}{\left(22414\times \frac{519.67}{491.67}\right)}}\times {\displaystyle \frac{18.0152}{453.59}}\right]+B$$

(13)

For some industrial applications, the expression of moisture in parts per million by volume of water, ${ppm}_{v }{H}_{2}O$, can be quite useful, as showcased in Equation (14):

$${ppm}_{v }{H}_{2}O={\displaystyle \frac{{Pv}_{ice}\times {10}^6}{{Pressure}_{PA}}}$$

(14)

To highlight the innovations of this study, we emphasize the comparative advantages of the proposed method over existing approaches. The temperature range of the ASTM D1142 test method, approach 1, is limited from 0 °F to 100 °F, which is completely different from the Brazilian real samples of natural gas. The ASTM D1142 test method, approach 2, simplifies the final algorithm to calculate the water vapor content, in addition to not being appropriate where the dew point is measured in conditions close to the critical temperature of gaseous fuels. These limitations are not present in the method used by the Portable Moisture Analyzer—PM880.

2.4. Measurement Uncertainty

The procedure proposed by JCGM 100:2008 [17] for estimating the measurement uncertainty of a given measurand consists of the following steps: definition of the measurand and its mathematical model, identification of the most relevant sources of uncertainty; quantification of these sources of uncertainty; reduction of uncertainties to standard deviation; combination of standard uncertainties and their respective sensitivity coefficients; declaration of the expanded uncertainty, coverage factor, and degrees of freedom for an appropriate confidence level, 95.45%, used in this work. The combined standard uncertainties can be calculated by Equations (15)–(22):

$$u_c^2\left({Pv}_{water}\right)={\left({\displaystyle \frac{12888759281487\exp \left(\frac{8751\times \mathit{DP}}{500\times \left(\mathit{DP}+\frac{24097}{100}\right)}\right) }{500{\times \left(100\times DP+24097\right)}^2}}\times u\left(DP\right)\right)}^2 DP>0$$

(15)

$$u_c^2\left({Pv}_{ice}\right)={\left({\displaystyle \frac{941902265421\exp \left(\frac{11271\times \mathit{DP}}{500\times \left(\mathit{DP}+\frac{6837}{25}\right)}\right) }{400{\times \left(25\times DP+6837\right)}^2}}\times u\left(DP\right)\right)}^{2}DP\le 0$$

(16)

$$u_c^2\left(IGT8\right)={\left({\displaystyle \frac{1711\mathrm{l}\mathrm{n}(10)\times {10}^{\frac{6687}{1000}-\frac{1711}{DP+\frac{5463}{20}}}}{{\left(DP+\frac{5463}{20}\right)}^2}}\times u\left(DP\right)\right)}^2$$

(17)

$$u_c^2\left(B\right)={\left({\displaystyle \frac{Pv}{{Pv}_{water}}}\times u\left(IGT8\right)\right)}^2+{\left({\displaystyle \frac{IGT8}{{Pv}_{water}}}\times u\left(Pv\right)\right)}^2+{\left(-{\displaystyle \frac{IGT8\times Pv}{{Pv}_{water}^2}}\times u\left({Pv}_{water}\right)\right)}^2$$

(18)

$$u_c^2\left({Pressure}_{PA}\right)={(33775000/4899\times u(P{ressure}_{psig}))}^2$$

(19)

$$u_c^2(X_{water})={\left({\displaystyle \frac{1}{{Pressure}_{PA}}}\times u\left(Pv\right)\right)}^2+{\left(-{\displaystyle \frac{Pv}{{Pressure}_{PA}^2}}\times u\left({Pressure}_{PA}\right)\right)}^2$$

(20)

$$u_c^2\left(W\right)={\left({\displaystyle \frac{{10}^6\times {30.48}^3}{\left(22414\times \frac{519.67}{491.67}\right)}}\times u\left(X_{water}\right)\right)}^2+{\left(u\left(B\right)\right)}^2$$

(21)

$$u_c^2({ppm}_{v }{H}_{2}O)={\left({\displaystyle \frac{{10}^6}{{Pressure}_{PA}}}\times u\left({Pv}_{ice}\right)\right)}^2+{\left(-{\displaystyle \frac{{Pv}_{ice}\times {10}^6}{{Pressure}_{PA}^2}}\times u\left({Pressure}_{PA}\right)\right)}^2$$

(22)

2.5. Measurement Uncertainty in Compliance Evaluation

Based on the information about the measurement uncertainty, it is possible to verify whether the results obtained are within the specified value. Consequently, it is essential to have a decision rule that considers the risks associated with an incorrect decision, which is determined by guard bands, g, that define acceptance and rejection regions (Figure 2).

Based on JCGM 106:2012, the guard band is established to ensure that the probability of a false acceptance or rejection occurring is less than or equal to a previously defined confidence level α when a measurement occurs in the acceptance region [18].

Each measurement result’s standard uncertainty, u, is added to the established guard band value. According to the Eurachem guide [19], a value of $1.64\times u$ is determined at a significance level of 5%.

In this study, we considered the risk of the consumer, which is defined as the probability of accepting a batch that ought to be rejected.

The histograms illustrated the most likely value, associated uncertainties, guard bands, and upper acceptance limits. To assess the risk to consumers, Monte Carlo simulations were generated from 100,000 pseudorandom values, ensuring a stability of 0.001%, for the moisture content in natural gas based on the mean experimental value and its expanded uncertainty.

Here, it was calculated using the MS-Excel function “=NORM.INV(RAND();$y_i$;$u_{y_i}$)”, where ${\mathit{y}}_{\mathit{i}}$ is the mean of the measured value and $u_{y_i}$ is its expanded uncertainty for the $i$th parameter, considering a Gaussian or normal distribution.

In recent decades, several studies have involved calculating uncertainty in the physicochemical properties of natural gas. Using measurement uncertainty, Oliveira demonstrated the extent to which the variation in molar fractions obtained from these two methods metrologically influences the higher calorific value of natural gas. Additionally, it affects the compressibility factor of natural gas [20]. Brazilian researchers compared five approaches to assess the uncertainty of natural gas composition by gas chromatography [21].

However, the use of measurement uncertainty in assessing the conformity of these parameters in natural gas is still very incipient. This is apart from a recent study where acceptance thresholds were established to ensure no risk of incorrect compliance in the carbon isotopic analysis of natural gas based on this uncertainty data [22]. On the other hand, this approach can be highlighted in other areas of knowledge, such as environmental pollution [23], drug and medicine analysis [24], and microbiology [25].

3. Results and Discussion

This section is divided into three parts. The introductory part compares the results from the Section 2.1, Section 2.2 and Section 2.3. approaches. The second part addresses the calculated uncertainties, while the third discusses using these uncertainties in conformance assessment. Section 3.2 compares the results from the Portable Moisture Analyzer (PM880) with the calculations described in Section 2.3.

3.1. Comparing Different Approaches

Although the objective of this study is not to compare different approaches for calculating moisture in natural gas, we compared three methods described in Section 2.1, Section 2.2 and Section 2.3. for determining the water vapor content based on dew point temperature measurements to assess any significant disparities.

We considered the water dew point to be 37 °F at a pressure of 15.0 psia as input data. After converting the temperature, 2.7778 °C, and the pressure, 0.304055 psig, the water vapor content could be calculated in pounds of H₂O/million standard cubic feet of natural gas, at a base of 60 °F and 14.7-psia pressure, Table 1.

Table 1. Comparing the approaches from Section 2.1, Section 2.2 and Section 2.3.

ASTM D1142 Test Method, Approach 1	ASTM D1142 Test Method, Approach 2	Portable Moisture Analyzer—PM880
342.8 lbs H2O/MMSCF NG	346.1 lbs H2O/MMSCF NG	345.6 lbs H2O/MMSCF NG

Table 1. Comparing the approaches from Section 2.1, Section 2.2 and Section 2.3.

ASTM D1142 Test Method, Approach 1	ASTM D1142 Test Method, Approach 2	Portable Moisture Analyzer—PM880
342.8 lbs H2O/MMSCF NG	346.1 lbs H2O/MMSCF NG	345.6 lbs H2O/MMSCF NG

In addition to the results of the approach in question, PM880, among others, the maximum difference found between them is 1%. This is a very interesting scientific result when considering the possible variabilities of each approach.

3.2. Uncertainty Evaluation

Between 13 September 2024 and 20 September 2024, 10 dew point (°C) and pressure (bar) results of natural gas from the Portable Moisture Analyzer—PM880 were collected to validate the algorithms of the developed MS-Excel spreadsheet. After the pressure data were converted to psig, the MS-Excel spreadsheet developed, which is attached as Supplementary Material in this study, calculated the moisture values and their respective expanded uncertainties, as shown in Table 2 and Table 3. In the spreadsheet, the input quantities in yellow, cells H2 and H3, are the dew point in Celsius degree and the pressure in psig, respectively, of the natural gas.

Table 2. Moisture results and their respective expanded uncertainties in lbs H₂O/MMSCF NG.

Dew Point (°C)	Pressure (psig)	Moisture Calculated by PM880 *	Moisture Calculated by the Spreadsheet *	Moisture Expanded Uncertainty Calculated by the Spreadsheet *
−27.4	1165.91	0.70	0.70	0.14
−38.0	455.37	0.41	0.42	0.08
−47.5	436.37	0.17	0.16	0.04
−32.1	555.30	0.64	0.65	0.12
−23.1	1129.80	1.01	1.02	0.19
−25.3	1121.384	0.84	0.85	0.16
−24.5	1113.99	0.91	0.91	0.17
−30.9	888.31	0.57	0.58	0.11
−41.0	443.62	0.31	0.31	0.06
−39.8	848.28	0.26	0.26	0.06

* lbs H₂O/MMSCF NG = pounds of H₂O/million standard cubic feet of natural gas.

Table 2. Moisture results and their respective expanded uncertainties in lbs H₂O/MMSCF NG.

Dew Point (°C)	Pressure (psig)	Moisture Calculated by PM880 *	Moisture Calculated by the Spreadsheet *	Moisture Expanded Uncertainty Calculated by the Spreadsheet *
−27.4	1165.91	0.70	0.70	0.14
−38.0	455.37	0.41	0.42	0.08
−47.5	436.37	0.17	0.16	0.04
−32.1	555.30	0.64	0.65	0.12
−23.1	1129.80	1.01	1.02	0.19
−25.3	1121.384	0.84	0.85	0.16
−24.5	1113.99	0.91	0.91	0.17
−30.9	888.31	0.57	0.58	0.11
−41.0	443.62	0.31	0.31	0.06
−39.8	848.28	0.26	0.26	0.06

* lbs H₂O/MMSCF NG = pounds of H₂O/million standard cubic feet of natural gas.

Table 3. Moisture results and their respective expanded uncertainties in ppmv H₂O.

Dew Point (°C)	Pressure (psig)	Moisture Calculated by PM880 *	Moisture Calculated by the Spreadsheet *	Moisture Expanded Uncertainty Calculated by the Spreadsheet *
−27.4	1165.91	6.08	6.1	1.3
−38.0	455.37	2.40	5.0	1.1
−47.5	436.37	1.64	1.7	0.40
−32.1	555.30	7.47	7.8	1.7
−23.1	1129.80	9.42	9.7	1.9
−25.3	1121.384	7.63	7.8	1.6
−24.5	1113.99	8.32	8.5	1.7
−30.9	888.31	5.40	5.6	1.2
−41.0	443.62	3.46	3.6	0.90
−39.8	848.28	2.14	2.2	0.50

* ppmv H₂O.

Table 3. Moisture results and their respective expanded uncertainties in ppmv H₂O.

Dew Point (°C)	Pressure (psig)	Moisture Calculated by PM880 *	Moisture Calculated by the Spreadsheet *	Moisture Expanded Uncertainty Calculated by the Spreadsheet *
−27.4	1165.91	6.08	6.1	1.3
−38.0	455.37	2.40	5.0	1.1
−47.5	436.37	1.64	1.7	0.40
−32.1	555.30	7.47	7.8	1.7
−23.1	1129.80	9.42	9.7	1.9
−25.3	1121.384	7.63	7.8	1.6
−24.5	1113.99	8.32	8.5	1.7
−30.9	888.31	5.40	5.6	1.2
−41.0	443.62	3.46	3.6	0.90
−39.8	848.28	2.14	2.2	0.50

* ppmv H₂O.

To calculate the expanded uncertainty of moisture, in the conservative approach, the expanded uncertainties of 2.0 °C, cell I2, and 1% of the full scale (300 to 3000 psig), cell I3, were considered as input data for dew point temperature and pressure, respectively, considering a coverage factor, k = 2. These conservative values of uncertainty sources already encompass other contributing factors, such as fluctuations in ambient temperature and moisture and long-term sensor drift.

The moisture results calculated by the developed spreadsheet are compatible with those generated by the portable moisture analyzer PM880 regarding the unit in lbs H₂O/MMSCF NG; however, it presents some discrepancies regarding the ppmv H₂O unit. The expanded uncertainty results, k = 2, with a confidence level of 95.45%, are in the range of 20%.

Therefore, a significant advantage is that the manuscript explicitly presents the algorithms, whereas commercial software operates as a “black box” without demonstrating the steps, calculating measurement uncertainty, or assessing compliance.

3.3. Use of the Information on the Measurement Uncertainty in Compliance Evaluation

In the MS-Excel spreadsheet, only the yellow cell H6 corresponding to the upper specification limit of 60 ppmv H₂O, which the Brazilian oil and gas industry is currently considering (Figure 3), needs to be filled in.

The guard band was calculated considering a significance level of 0.05 and an upper specification limit, USL. USL calculated the upper tolerance limit for the moisture in natural gas minus $1.64\times u$.

Figure 3 provided the histogram for the measurement value, the specification, and the guard band limit, considering p(AU)—probability density at the upper acceptance limit, AU—upper acceptance limit, and TU—upper tolerance limit.

The natural gas’ measured moisture value, 9.7 ± 1.9 ppmv H₂O, was significantly below the upper tolerance limit, suggesting compliance with the specification. Consequently, this indicates an estimated consumer risk of 0.0% associated with this measurement, as the upper tolerance limit is 58.4 ppmv H₂O.

4. Conclusions

This study developed and validated an MS-Excel spreadsheet for assessing the uncertainty of natural gas moisture based on the Portable Moisture Analyzer—PM880 algorithms. In this preliminary study, the sources of uncertainty considered were the dew point temperature and the natural gas pressure.

Even with a measurement uncertainty value of around 20%, there is no risk of false compliance assessment, as the moisture levels in natural gas are significantly lower than the upper specification limit.

Regarding future research, it is important to evaluate the impact of sampling variability on measurement uncertainty, as this factor can influence risk assessment. Furthermore, the calculations for measuring the dew point and pressure could be elaborated upon by incorporating more realistic estimates of their uncertainties. The sensitivity analysis on the uncertainty sources (dew point temperature and pressure) could also be included in cases where measurement uncertainty can compromise the compliance evaluation. Another area for future work is comparing the results between different methods.