A 'recipe' for reliable gathering system modeling

Considerations

The following is a systematic guide to modeling gas gathering systems. It can be used as a recipe to decrease your time spent on modeling and increase the reliability of your modeling forecasts. It has been specifically designed to work with IHS Piper. Before outlining the methodology, there are several considerations that should be addressed.

• It is critical that the objectives of the modeling project be clearly defined and it is recommended that the model be constructed with just enough detail to attain the objectives. For example, assume that several new wells are going to be tied-in to an existing gathering system, and the impact on well backpressures and compression capacity needs to be understood. The model's detail should focus on the compressor capacity and understanding the reasons for the current pipeline pressure losses. Although useful and perhaps even informative, modeling the wellbores is not required to attain the prime objectives and will add considerable time to the project work.
• Gather gas rates and flowing pressure data from daily operations sheets, SCADA, or other forms of remote measurement. This data should include wellhead tubing and casing pressures, well line pressures, as well as compressor suction and discharge pressures. Gather at least two weeks of this data and scan it for a relatively stable period. If the system is continually unstable, then you have already made an important observation that requires follow-up! Once you pick a stable point, which we will call our "match point," sum the gas rates for all wells and compare them to the reported delivery point gas rates. They should be reasonably close to each other.
• There is sometimes confusion over the need to match a gathering system model over the life of the project or match a point in time. Reservoir analysis methods for determining hydrocarbons in-place need to incorporate the entire production history whereas surface system analysis should be conducted at a point-in-time. Consequently, any gas gathering modeling study should include or be prefaced by a comprehensive reservoir analysis that produces a determination of well groupings and hydrocarbon-in-place for each reservoir accumulation. On the one hand, trying to historically model the gathering system is not recommended because the surface system is almost always in a state of flux-with new wells being added, older wells going on and off line, compressors being added or enhanced, and piping being reconfigured. It is often difficult to determine the current operating parameters, and almost impossible to figure out the system configuration that occurred six months, two years, five years, or twenty years ago.
• Use a compartmentalized methodology to break the work down into manageable parts. We recommend the four-step process, as outlined in this article.

STEP 1: Pipeline Pressure Loss Match

The pipeline pressure loss match focuses on the pressure losses in the surface pipeline system only.

1. Use Fixed Gas Rate wells to set the gas rate for each well to the rate measured at the match point. Attaching a pool is optional. Generally, reservoir data is not usually available during this phase of modelling projects.

2. Set-up compressors to operate at a fixed suction pressure equal to the match point suction pressure. In most situations, you should not restrict flow capacity or horsepower at this stage of the process.

3. Set-up the system so you can easily monitor the calculated flowing pressures versus the match point (measured) flowing pressures. Start at the delivery point and work sequentially outwards from measured pressure point to measured pressure point, comparing calculated with measured values. The delivery point pressures and compressor suction pressures should match exactly because they are "givens."

4. Conduct the preliminary model match. Resist the temptation to use tuning factors because these parameters can easily create inappropriate pressure loss sensitivities at higher or lower gas rates. Instead, use the model to identify problem areas and create a list of issues that require resolution. Keep the following points in mind while attempting to reconcile the model calculations and field measurements.

• There are always errors in measurements. We recommend starting with an allowable pressure difference of 20 psi (150 kPa) for matching. As you gain confidence in the measurements, decrease the allowable pressure differential. If you find problems with the quality of the measurements, you may need to increase the allowable pressure differential. Generally, the accuracy and quality of recorded pressure measurements increases with decreasing pressure. This is not strictly a gauge accuracy issue. We find that operators pay more attention to pressure measurements as system pressures are lowered.
• Categorize pressure loss differences (measured versus calculated) as gradual or step-change. In the case of gradual pressure loss differences, you can expect the differential between measured and calculated pressures to become greater as you work your way out a pipeline leg. In the case of step-changes, the excess pressure loss can be attributed to a localized portion of the pipeline. Gradual pressure changes are due to a systemic modeling problem; missing wells, running single-phase when a two-phase correlation should be used, and so on. Step changes are usually due to localized events such as stagnant liquids (non-moving liquids) at a river crossing, plugged filters, undersized headers, undersized valves, and so on. A detailed explanation of stagnant liquid columns and their impact on gas gathering systems can be found in "An Effective Method for Modeling Stagnant Liquid Columns in Gas Gathering Systems." This paper (2004-175) was presented at the Petroleum Society's 5th Canadian International Petroleum Conference (55th Annual Technical Meeting) in Calgary, Alberta, Canada, June 8 10, 2004.
• Categorize step-change pressure losses as frictional or hydrostatic. Frictional pressure losses are usually caused by an undersized system component whereas hydrostatic pressure losses are almost always stagnant liquids. In rare occurrences hydrostatic pressure losses can be due to extreme elevation changes, such as traversing up or down a mountain.

5. Go to the field prepared to take independent pressure measurements. As each issue is resolved, adjust the model to incorporate the change. Resolving one issue often fixes several problems. Continue the field trip until all issues are resolved! This will be your "pipeline pressure matched model."

A portion of a matched model is included as Fig. 1. The delivery pressure at the 06-06 Plant was set at 500 psia. The label for each well includes (from top to bottom) the well name, calculated pipeline pressure, gas flowrate, and measured pipeline pressure. Comparison of the calculated and measured line pressures demonstrates that an excellent match was achieved. This match required the inclusion of a step pressure loss that was interpreted to be a "Stagnant Liquid Column" in a river valley. The gas velocity of 0.76 ft/s in the river valley indicated that two-phase flow in the uphill section of the valley was unlikely.

STEP 2: Well Deliverability Match

Step 2 involves selecting a well deliverability option. Depending on the complexity of the model and the objectives, deliverability can be modeled at the wellhead or the sandface. In either case, the final deliverability match must closely model current deliverability from each well in the system. Key parameters to review/adjust, without a match, are sandface C and n, current reservoir pressure, and pressure loss correlation (single-phase as opposed to multi-phase). Use of tuning factors should be avoided.

Sandface and wellhead C and n values are often subject to poor testing and analysis practices. Current reservoir pressure depends heavily on inputs of original gas-in-place, initial reservoir pressure and cumulative gas production. Any errors in these values can adversely affect the current deliverability estimate.

An example of deliverability modeled at the sandface is included as Fig. 2. The upper curve of black triangles is the sandface deliverability.

The lower curve of mauve diamonds is the wellhead deliverability curve as calculated from the sandface deliverability curve and the wellbore tubulars description. If all the data fits together, the current test conditions (represented by the red dot) should land on the wellhead deliverability curve. If not, model forecasts will not match current operating conditions.

STEP 3: Compressor Match

Immediately upon receipt of compressor capacity curves, plot the current operating conditions (Gas Rate and Suction Pressure) on the compressor capacity curve! The current rate and suction pressure should appear on or very close to the capacity curve. If not, find out why. Compressor capacity curves should only be used in a model after they have been validated with current operating conditions. This is extremely important. Mismatches between field deliverability and compressor capacity are one of the most common reasons why models have trouble replicating actual field conditions.

Compressors are often run at different conditions than the "supplied" capacity curves. Capacity curves most often reflect original design parameters. Operators will sometimes run the units at a slightly lower rotating speed (RPM) to reduce downtime for scheduled maintenance. Valve pockets, valve spacers, rotating speed, and cylinder unloading are strategies often used for capacity control (match compressor capacity to field deliverability). Compressor discharge pressure can also vary greatly. If any of these situations occur, it is important to get capacity curves that more closely reflect the current operation. Revised curves can usually be obtained from the original manufacturer or packager. If this is not possible, we recommend that you use a compressor performance program to match the design curve, and then generate new performance runs based on current conditions. Revising compressor curves is not recommended unless you have specific knowledge in this area. A matched set of capacity curves is shown in Fig. 3.

STEP 4: Base Case Run and Gas Rate Forecast to History Match

Run the base case forecast with the finalized model: final pipeline pressure match, validated deliverability curves and validated compressor curves. For each well in the system, obtain a plot of gas rate production history and plot the model-generated production forecast on it. The current gas rate should line up closely to recent history. The forecast decline trend should match closely with the historical decline trend. If the rate does not match, then the match point probably used a high or low fluctuation in flow rate. The deliverability will need to be revised to reflect the average gas rate for this well. If the forecasted decline trend does not match, then the reservoir original gas-in-place probably needs to be revisited. If the forecast trend slope is too shallow, then the OGIP was most likely overstated. If the forecast trend slope is too steep, the OGIP was most likely understated.

Additional considerations:

Ralph McNeil, Senior Technical Advisor, IHS Energy
Posted November 19, 2015

This article was published by S&P Global Commodity Insights and not by S&P Global Ratings, which is a separately managed division of S&P Global.