1

PAPER 2005-230

Gas Gathering System Modeling The Pipeline Pressure Loss Match

R.G. MCNEIL, P.ENG. Fekete Associates Inc.

D.R. LILLICO, C.E.T. Fekete Associates Inc.

This paper is to be presented at the Petroleum Society’s 6th Canadian International Petroleum Conference (56th Annual Technical Meeting), Calgary, Alberta, Canada, June 7 – 9, 2005. Discussion of this paper is invited and may be presented at the meeting if filed in writing with the technical program chairman prior to the conclusion of the meeting. This paper and any discussion filed will be considered for publication in Petroleum Society journals. Publication rights are reserved. This is a pre-print and subject to correction.

Abstract Creating a gas gathering system model that is capable of

accurately reproducing current rates and backpressures, as well as being able to predict future operating conditions after adding new wells, installation of a loop or booster compression can be a challenge. It does not need to be difficult though, as long as the modeler takes a comprehensive, logical approach. The approach required begins with three overriding rules: 1) break the problem into manageable pieces, 2) select an appropriate pressure loss correlation and trust it without utilizing any adjustment factors, and 3) always conduct a field trip to resolve differences between measured and calculated pipeline pressure losses.

The objective of this paper is to present the modeling

approach developed and utilized by our pipeline modeling group called the “Five Step Modeling Method”1. This is the first of several papers that will describe in detail, with cases studies, each of the modeling steps; 1) Pipeline Pressure Loss Match, 2) Well Deliverability Match, 3) Compressor Capacity Match, 4) Base Case, and 5) Production History to Forecast Match. This paper presents the Pipeline Pressure Loss Match.

Introduction This paper is based on a gas gathering system operated by a

Canadian company in Central Alberta. Currently, natural gas from a total of 16 wells is produced to a compressor station and then sent to sales. The purpose of the model is to simulate the effect of the tie-in of a number of low-pressure wells.

Since the construction of the pipeline links portion of model

is largely a mechanical operation, discussion will focus on the process of gathering, interpreting and utilizing field performance data to match measured pressure losses to modeled pressure losses.

Application of the pipeline pressure loss match begins with

the gathering of performance data and selection of a match point. Most gas gathering systems are run in a constant state of flux; wells are produced intermittently, facilities temporarily go offline for a variety of reasons, gas is diverted to another system or compressor, and new wells or facilities are added. Consequently, it is very difficult to match systems over extended periods of time and so most models are matched at a

PETROLEUM SOCIETY CANADIAN INSTITUTE OF MINING, METALLURGY & PETROLEUM

2

point in time. Once the match point has been selected, the performance data is compared to the model calculated data. The key from this point onward is to highlight the differences between the measured and calculated data and gather additional data to resolve those differences. Experience has taught us that the differences are usually not deficiencies in the model but rather unknown factors in the field. As a result, field trips have become an integral part of the modeling matching process.

Performance Data The performance data required to match a model to current

operating conditions are 1) the wellhead and line pressures plus the current flowrate (gas and liquid) for each well, 2) suction and discharge pressures plus throughput at each compressor. This information is typically gathered on at least a daily basis and is generally readily available. At least several weeks to a month should be collected and reviewed prior to selection of the match point. Figure 1 presents a sample data set for a well from December 1, 2004 to February 24, 2005.

In Figure 1, the well does not produce gas from approximately December 20, 2004 to January 25, 2005 and so this period of time is not desirable for selection as the match point. The flowing tubing pressure data is relatively continuous, meaning that for most days there is a matching tubing pressure for the daily gas flowrate. The line pressure, which is the critical flowing pressure required for the pipeline pressure loss match, is not continuous but fortunately it corresponds closely to the more readily available tubing pressures. Note all measured pressure data are plotted using symbols with a line connecting the points regardless of missing data. This was done to make it easier to see the pressure trend.

There are three main strategies for selection of data for the

match point. The first strategy is to pick a point-in-time that maximizes the number of wells on-stream and the number of available data points. This method is desirable when retrieving large data sets from databases or spreadsheets and works well when the data set is complete and operating conditions are stable. It does not work well when flowrates and pressures are variable or when there are long periods of missing data as illustrated with the line pressures in Figure 2. The data points are extracted from specific measurement points shown graphically in Figures 1 and 2; the intersection of the dashed vertical line (indicates match point date) and the plotted data point (gas flowrate and line pressure).

The second strategy requires picking of average values

based on the trend seen in the data. The advantages of this method are that it eliminates selection of “peak or valley” values and allows the modeler to “fill-in” missing data. This method is shown in Figures 1 and 2; the intersection of the dashed vertical and the dashed horizontal lines indicates the selected data points. The major downfall of this strategy is the tendency to apply it to cases where the data is too sparse.

The third strategy is to break the pipeline system into

sections and conduct a model match on each of the sections usually at different match point dates. This style is generally used when gas flowrates are metered continuously only at a group point and flow in the individual pipelines are only periodically metered during a test period. Matching of those pipeline segments must be conducted using the test data at the point in time that it was tested. The main superstructure of the system, from the group meters to the delivery point, can be handled using the more conventional strategies outlined above.

In this model, many of the wells exhibited similar performance data to that of the 04-17 well as shown in Figure 2. Consequently, the second strategy was used select the match point data. A comparison of the match point selection data for the first and second strategies is presented in Table 1. The match point was selected to be February 18, 2005. The data in Table 1 shows the selected performance data can differ greatly.

Although the line pressures measurements are sparse for the

04-17 well, they tend to mirror the tubing pressures very closely. Therefore, if no line pressure had been taken for the match point date, the average line pressure value could have been selected partially using the tubing pressure data.

Model Setup Model setup is kept very simple at this stage. Gas flowrates

were entered as “fixed” values and compressors are entered as simple devices with a set suction pressure equal to the match point measured suction pressure. This approach is taken so that all attention can initially be focused on the pressure losses in the pipeline system. Consequently, the pipeline pressure loss match can proceed before any information about the well deliverability or compressor capacity is collected.

The recommended practice for matching measured and

calculated pressure data is to first look at the variance in the line pressure measurements. If sufficient data is not available, then it is recommended that an error difference of 20 psi or 150 kPa be used. In this case, it was determined that a 20 psi error difference would be used, initially.

Map 1 presents the preliminary model run using the selected

performance data. The historical performance data indicated there was some water and condensate produced by the wells but the volumes were not considered accurate and most of the pipelines traverse is slightly downhill to the plant. Therefore, the model was initially run using a single-phase gas pressure loss correlation.

In the system maps the displayed information includes the

well name, the calculated line pressure (PL), the gas flowrate (qg), and the measured data shown as “WH” for the wellhead tubing pressure and (L) for the line pressure.

Preliminary Model Match The preliminary model match is presented as Map 1.

Starting at the plant and working sequentially outwards, it can be seen that the line pressures match reasonably for wells near to the plant but progressively worsen further from the plant. This pattern is categorized as a systemic pressure loss problem; the pressure loss error gradually cumulates with distance from the delivery point.

After consulting with the operator, it was determined that

most of the wells have on-site 3 phase separation that removes the water and condensate but the condensate is then re-injected back into the line. Some of the wells have no separation and so are produced directly to the pipeline. Some wells have all produced fluids trucked offsite. The operator was also able to provide additional data on the volume of liquids production at each wellsite. The model was subsequently enhanced to include elevation changes, liquid volumes and switched to a two-phase

3

pressure loss correlation. The resulting model is presented as Map 2.

Starting at the plant and moving outward to the wells

nearest the plant, it can be seen that the calculated pressures match very closely with the measured line pressures. This indicated that there is very little pressure loss from the compressor suction to the plant inlet header.

Working further outwards from the plant in Map 2, the

match for the 05-10, 03-10, 04-16, and 04-17 wells was within 4 psi. As confidence in the pressure data increased, it was decided that the error pressure difference could be reduced to 10 psi (70 kPa).

Working further upstream, it was noted that gas velocities

decreased and several potential fluid traps were identified from elevation mapping. Creek crossings were identified in the vicinity of the “Jct 14-26” and “Jct 15-03” nodes in the system and a dip in the pipeline was also noted in the vicinity of the “Jct 12-02” node. It has been found that a common cause of excess pressure loss in gas pipeline systems is the buildup of non-moving liquids in liquid traps called Stagnant Liquid Columns2. As discussed in previous work, these fluid buildups should be modeled using a fixed pressure loss facility, as they are typically very stable once established. Therefore, a fixed pressure loss of 10 psi was added to the Jct 12-02 node and 10 psi to the Jct 14-26 node. These additions improved the match in the 06-07, 12-02, 01-32, 03-34 and 04-04 wells such that all calculated line pressures were now within 9 psi of the measured line pressures. The error differences for the remainder of the wells, except for the 03-06 well, were within the 10 psi error limit. The 03-06 well was off by 23 psi. Although a creek crossing was identified at the Jct 15-03 node, a fixed pressure loss was entered at this location because there was no upstream gas production. The preliminary model match including elevations, liquids, two-phase pressure loss correlation and fixed pressure losses to model non-moving liquids is presented as Map 3.

In spite of the apparent good match, there was considerable

“fill-in” or trending of the measured data to extract line pressures. Therefore, it was decided that a field trip was necessary to gather additional pressure data and understand better the field operation.

Field Trip The field trip was conducted on March 21, 2005. A

meeting was held with the field operator and updated information in the form of current well flowrates was obtained and used to update the model. In addition, it was learned that the 07-12 well had been added to the system and a small compressor was on on-site to help maximize well productivity.

The plant compressor setup was reviewed and several

confirming pressure measurements were obtained. In addition, several wells along the main line were visited and pressure measurements were obtained. It was concluded during the field trip that the quality of the field measured pressure data is very good. It was also concluded that the creek crossing at Jct 15-03 requires a fixed pressure loss of 15 psi based on pressure measurements associated with the new 07-12 well.

Finalized Model Match The finalized model match is presented in Map 4. Beyond

the addition of three fixed pressure losses to model the buildup of liquids, the model predicts the pipeline pressure losses within acceptable error limits.

The finalized model indicates that running the system in two-

phase flow as opposed to single-phase gas flow increases well backpressures by an extra 10 psi for the wells nearest to the plant compressor to approximately 60 psi at the margins of the system.

The 4 inch pipeline noted in Map 5 is currently operating at a

gas velocity of 36 ft/s and a frictional pressure loss of 25 psi/mile. The 6 inch pipeline noted in Map 5 is currently operating at a gas velocity of 39 ft/s and a frictional pressure loss of 12 psi/mile.

A trial scenario was run to investigate the addition of 5

MMscfd at the 03-34 location. It was assumed that sufficient compressor capacity would be added to maintain the suction pressure at 70 psia. This case demonstrates that an additional 5 MMscfd will result in increased well backpressures ranging from 40 to 160 psi. Map 6 presents the results.

Summary The case presented herein demonstrates that much can be

learned about the operation of a gas gathering system if the simulation model is utilized to highlight and understand the system dynamics rather than force fitting the pressure loss correlation to match a data set. This understanding can then be carried through to a simplified scenario stage that can very simply answer some key questions.

Acknowledgement We would like to acknowledge the help and advice

provided by David Penton of Samson Canada Limited in completing this paper.

NOMENCLATURE qg = Gas Flowrate, MMscfd qL = Liquid Flowrate, Bbl/d qgI = Compressor Inlet Flowrate, MMscfd PL = Line Pressure, psia Pd = Delivery Point Pressure, psia Psuc = Compressor Suction Pressure, psia PO = Compressor Discharge Pressure, psia ID = Pipeline Internal Diameter, inches vg = Superficial Gas Velocity, ft/s FL = Pressure loss due to Friction, psi/mile

REFERENCES 1. McNeil, R.G., “Video 5, A 5 Step Approach To

Modeling Gas Gathering Systems”, Fekete Associates Inc. Technical Video Presentation, April 2005

2. McNeil, R.G., Lillico, D.R., “An Effective Method for Modeling Stagnant Liquid Columns In Gas Gathering Systems”, Presented at the Canadian International Petroleum Conference, Calgary, Alberta, June 2004

4

03-06 Well

Daily Performance Data

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

12/1

/04

12/8

/04

12/1

5/04

12/2

2/04

12/2

9/04

1/5/

05

1/12

/05

1/19

/05

1/26

/05

2/2/

05

2/9/

05

2/16

/05

2/23

/05

Gas

Rat

e, M

Msc

f

0

200

400

600

800

1000

1200

1400

Pre

ssu

re, p

sia

Gas RateTubing PressureLine Pressure

Figure 1 - Daily Performance Data

04-17 WellDaily Performance Data

0.0

0.1

0.2

0.3

0.4

0.5

12/1/04

12/8/04

12/15/04

12/22/04

12/29/04

1/5/05

1/12/05

1/19/05

1/26/05

2/2/05

2/9/05

2/16/05

2/23/05

GasRate,MMscfd

0

50

100

150

200

250

Pressure,psia

Gas RateTubing PressureLine Pressure

Figure 2 - Daily Performance Data

5

Strategy 1 Strategy 2Gas Line Gas Line

Well Flowrate Pressure Flowrate PressureMMscfd psia MMscfd psia

03-06 1.100 No Data 1.100 200

04-17 0.420 110 0.360 110

Table 1

6

Map 1 – Preliminary Model Match – No Liquids

7

Map 2 – Preliminary Model Match – With Liquids

8

Map 3 – Preliminary Model Match – With Liquids & Fixed Pressure Losses

9

Map 4 – Finalized Model Match – With Liquids & Fixed Pressure Losses

10

Map 5 – Finalized Model Match – Areas of Concern

11

Map 6 – Scenario 1: Add 5 MMscfd at the 03-34 Location

12