Parameters Tab (Probabilistic & Sensitivity Analyses)

The Parameters tab is where you define the distribution types and characteristics of the inputs for your probabilistic analysis. You can also define the number of runs used in the simulation and view the associated distributions of each input.

The table at the bottom of the tab lists all of the inputs to which you are able to assign uncertainty.

The values used for each parameter come directly from the model you have linked to the worksheet. The Distribution drop-down menu options are as follows:

Constant	Triangular	APE
Normal	Uniform	Dependent (only available for saturations)
Log-Normal	Estimate

Note:   Some parameters may show Not used; this means that they are optional parameters that were not used in the linked model.

If the initial water saturation is not 100%, then C_gi is set to the max value for C_gi because the PSS model needs the coal to be saturated for the calculations to work. Unlike conventional gas wells, gas in the pores of an unconventional CBM well accounts for very little of the total gas. (You can see in a CBM volumetrics analysis how little gas is in the pores.)

If you click the Cgi Options button, the Gas Content (GC) Options dialog box opens.

This dialog box is only used for layers with 100% water saturation. When performing a risk analysis for V_L, P_L, C_gi, or initial pressure, the resulting combination often results in an impossible combination of values. By default, these runs are discarded. However, you may change C_gi or V_L to make an acceptable combination of values.

Prevent over-saturation. Over-saturation occurs when C_gi is above the Langmuir isotherm. Because C_gi must be on or below the isotherm, we can:

• Discard the run (deselected as shown above)
• Adjust the Langmuir isotherm up (keep the C_gi)
• Suppress gas content by reducing the value of C_gi (keep the isotherm)

Note:   The under-saturation lower limit increases the C_gi to the lower limit, as per your settings.

Ensure that there is saturation. If the coal is gas saturated and the isotherm line or initial pressure is moving, move the C_gi on to the line. If the GC options are set, the results are less random. However, you must first determine what input variable you want randomized (that is, do not suppress gas content if C_gi is the random variable).

Constant

By default, the distribution type for all parameters is set to Constant — which means that no uncertainty has been applied.

Normal

By selecting Normal, you can apply a normal distribution to a particular parameter by entering a mean and variance, or a Min (P90) and Max (P10).

Log-Normal

By selecting Log-Normal, you can apply a log-normal distribution to a particular parameter by entering a mean and variance, or a Min (P90) and Max (P10).

Triangular

By selecting Triangular, you can apply a triangular distribution to a particular parameter by entering a minimum, mode, and maximum. The minimum and maximum values define the extent to which the probability density is equal to zero. The mode defines the peak.

Uniform

By selecting Uniform, you can apply a uniform distribution to a particular parameter by entering a minimum and maximum.

Estimate

If you are unsure of the distribution type of a particular parameter, Harmony can estimate one for you. (You must have a sample set of measurements to analyze.)

Selecting the Estimate option opens the Estimation Values dialog box.

You can either manually enter values, or paste your sample values into this dialog box. If you have more than 15 samples, the number of rows expands as required. Click OK to close the dialog box. Harmony estimates the parameters that fit your data.

Automatic Parameter Estimation

The automatic parameter estimation (APE) option is only available for parameters that have this option available in the linked model. Note that you must have a production history that consists of both rates and a calculated sandface, or active pressures in order to use APE.

APE attempts to minimize the difference between the synthetic curve and the historical data by modifying the selected parameter. This history-matching process occurs on each individual run during a probabilistic simulation, and thus it can take quite some time. Using APE can be very beneficial in optimizing probabilistic simulation results, and reducing the randomness of the process.

When APE is selected, an APE Options button becomes available. When you click this button, you can enter a maximum acceptable APE error. When this maximum error is specified, the APE error of each run is compared to this maximum error. If the maximum APE error is exceeded, that particular run is discarded.

For more information, see Automatic Parameter Estimation Theory.

Dependent

The Dependent option is only available for saturations. When a saturation is assigned a distribution, another saturation must be adjusted, so that the sum of all the saturations remains at 100%. By selecting a Dependent saturation, you specify which saturation is recalculated to satisfy these conditions, while the other saturation remains constant.

If oil saturation is set to zero, this parameter must be set to Constant to perform a probabilistic analysis. If oil saturation is non-zero, oil properties must be entered to perform a probabilistic analysis.