Layer | Fill | Outline |
---|
Map layers
Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|
Fill | Stroke |
---|---|
Collaborating Authors
Results
ABSTRACT Model calibration is the act (some might say "art") of adjusting model parameters in such a way that the model's behavior matches as closely as possible the behavior of the real-world system that it represents. In order to successfully calibrate a hydraulic model, certain hydraulic conditions must be known in order to have a defined calibration solution. Pipes that run parallel to each other (i.e. from the same upstream location to the same downstream location in roughly the same right-of-way) can pose serious difficulties to this requirement, especially when no inline flow measurement on any of the parallel lines exist, as the lack of knowing the exact flow distribution between the parallel lines means that the calibration problem either has no finite solution, or the finite solution is exceedingly difficult to determine. A potential solution to this problem involves utilizing multiple data sets. Each data set will have a particular range of possible solutions, and by comparing the solution ranges of multiple data sets, a single solution can easily be found. This paper will describe this method and provide examples with the intent of enabling the reader to apply the methodology to his or her own hydraulic calibration challenges. INTRODUCTION AND BACKGROUND Most engineers involved with hydraulic simulation are probably quite familiar (too familiar?) with the Darcy-Weisbach flow equation that describes head loss in terms of flow, pipe length, and pipe diameter. A form of the equation is shown below, as understanding the equation will be crucial to understanding the fundamental difficulty of calibrating parallel pipes.
ABSTRACT Calibration is an often-overlooked aspect of hydraulic modeling, but the impact of ignoring this potentially crucial step can be immense. It is not uncommon for significant pressure losses to be incurred in relatively small facilities that can often be glossed over as mere minutia when constructing a model. This paper will provide insight into the challenges faced while calibrating Access Midstream's Barnett hydraulic model and some of the solutions that arose during that process. There are a number of areas that can cause a model's hydraulic behavior to differ from that of the system it is attempting to represent. These areas include the following:Model Structure Model Integrity Measurement Data Integrity Hydraulic Integrity Flow Loop Handling INTRODUCTION Access Midstream is a publicly-traded Master Limited Partnership (MLP) that was spun off from Chesapeake Energy's midstream division in 2013. It has operations in 7 regions spread across 9 states with an average throughput of 3.8 billion cubic feet per day (bcf/d) and more than 6,700 miles of natural gas gathering pipelines. Access's Barnett assets include approximately 860 miles of gathering pipeline with a throughput of over 1 bcf/d. The system includes 24 compression facilities using more than 154,000 horsepower. The Barnett hydraulic model includes 615 receipt points (including Chesapeake well pads and interconnects with third parties) and 59 delivery points. For calibration purposes, data from over 2,500 meter stations are used, with most stations providing flow, pressure, temperature, and composition data.
- North America > United States > Texas > Fort Worth Basin > Barnett Shale Formation (0.99)
- North America > United States > Texas > Fort Worth Basin > Barnett Field > Barnett Shale Formation (0.98)
ABSTRACT The use of an efficiency factor to make pipe equations match physical reality is not a new concept - it's been around as long as there have been flow equations. It fell into some disrepute and was deemed unnecessary by some when sound theoretical equations for pipe flow began to replace the older empirical ones. Efficiency, however, is of more use that just fixing bad equations since it also is useful in adjusting specific pipes for problems and considering operational issues. A previous paper, A Tutorial on Pipe Flow Equations, presented at the 2001 PSIG meeting as a replacement paper in the wake of 9/11 but not published with the proceedings since it was too late, ended with the thought that pipeline efficiency was a valuable tool in calibrating gas models, more so than that of pipe roughness. Since then, I have received much verbal support from people within the industry but continue to hear comments that pipe roughness should be used as "the" tuning parameter. This paper builds on the original paper to explore the concept of pipe efficiency, its effect on flow equations, and its value as a calibration tool. Along the way some concepts regarding system design in the face of load variance within a day are also presented. Also some considerations with using the Panhandle equations that have been lost over time are mentioned. 1. Introduction and Problem Statement The flow of natural gas through pipes is well known in the literature and will not be re-derived here. For more details, please refer to the earlier papers referenced in the bibliography, particularly the excellent detailed derivation in the one by Susan Gibson from the 1981 PSIG conference. 2. Flow Equation Problems As stated above, all of the "practical" equations make some simplifying assumption about the variance of friction factor with flow ranging from constant values to explicit exponential functions. This gives rise to the fact that these equations are only valid within some range of conditions and must be corrected as conditions change. For example, my experience with the Weymouth equation has shown that at typical diameters around 20" and appropriate flows, efficiencies of 106% are often required to make the equation match observed data in a truly steady-state case. Since the forms based on the Moody Diagram surmount these problems, the remainder of this discussion, except for the following comments regarding the Panhandle equations, will deal only with the Colebrook-White equation, although its conclusions are equally valid for the GERG equation and its explicit forms. Therefore, this component of efficiency, e1, will be considered to be 100% or 1.00 since the equation should not need correcting. For those still using the Panhandle equations, either "A", "B", or some variant thereof, there is a further consideration that seems to have been lost in antiquity.
- North America > United States (0.46)
- Europe > United Kingdom (0.28)
ABSTRACT Pipeline simulation tools include steady state, transient and on-line computer programs. The determining factor on how useful a simulation tool is is its ability to predict a line's capacity under a wide range of conditions. Southern Natural Gas had a simple model to predict firm flows during winter conditions. However, the model did not produce accurate predictions under conditions that varied significantly from a peak winter day. Recognizing that our modeling limitations were mostly self-imposed, SNG has transformed its model into a truly useful tool. In order for a model to be versatile, many details must be taken into account. An understanding of the equations pertaining to flow is an important first step towards building a good model. Observations will be made concerning the various gas, pipe and environmental variables and how they impact the flow through a line. In addition, several methods of determining the friction factor will be examined, with the focus on Colebrook and AGA. This paper will present a hierarchal need of a model for accurate data. The criterion used in creating this hierarchy is the extent to which a tuned model compensates for errors in the data set over a wide range of conditions. As a useful model depends on proper tuning, several tuning techniques are presented. They include: Steady state tuning Steady state tuning with transient factors Transient tuning On-line tuning Comparisons of the different tuning methods will be presented from actual studies performed on the SNG system. They will be evaluated on the basis of how consistent the resulting pipe efficiencies are over several studies. Tuning via pipe efficiency and roughness will also be addressed. 2 Background Southern Natural owns and operates two pipeline systems in the southeast United States: Southern Natural Gas and South Georgia Co. Both are fully subscribed pipes that have a tariff provision that allows shippers to take gas at 6% of the daily quantity in any hour. Historically, winter was the only critical season as SNG served primarily heat sensitive loads. The fact that SNG was fully subscribed was moot as there was no summer market. As a result, the model was primarily a planning tool that was used to simulate flows during peak winter days. Operations would occasionally analyze the impact of a facility outage. However, flow studies were only a small part of their job function - due in part to the amount of time it took to conduct. Because of various assumptions and oversights, the model had to be tuned each time a study was conducted. Thus it provided little more benefit to operations than being a record base of pipe, compressor and connectivity information. With the growing power generation sector, SNG's world began to change. Loads in the summer were straining the system and maintenance windows were shrinking. As the Operational need for faster flow analyses became more prevalent, deficiencies in the model became intolerable.
- Production and Well Operations > Well & Reservoir Surveillance and Monitoring > Production logging (0.93)
- Facilities Design, Construction and Operation > Pipelines, Flowlines and Risers > Piping design and simulation (0.60)
- Facilities Design, Construction and Operation > Measurement and Control > Pipeline leak detection (0.60)