Expert Adviser Program for Rod Pumping (includes associated paper 19367)

抽油杆 测功机 人工举升 Lift(数据挖掘) 专家系统 领域(数学) 集合(抽象数据类型) 计算机科学 机械工程 工程类 人工智能 数据挖掘 石油工程 数学 程序设计语言 纯数学
作者
William L. Foley,J. G. Svinos
出处
期刊:Journal of Petroleum Technology [Society of Petroleum Engineers]
卷期号:41 (04): 394-400 被引量:19
标识
DOI:10.2118/16920-pa
摘要

Summary An expert adviser program for rod-pumping diagnostics (EXPROD) is being used successfully in the majority of oil fields operated by Chevron U.S.A. Field studies show that EXPROD gives excellent results in more than 90% of the cases analyzed. The program starts with results obtained from an analysis of field dynamometer data. These results consist of surface and downhole dynamometer cards, pumping-unit torque, rod stresses, and other parameters. From these data, EXPROD produces a one-to two-page diagnostic report in paragraph form that identifies equipment problems and recommends solutions. The problem-solving strategy used by EXPROD consists of a statistical pattern recognition scheme coupled with a network of rules based on company expertise in rod-pumping diagnostics. With this logic, EXPROD attempts to find the best diagnosis or a set of alternatives to account for a problem. This approach differs from the widely used rule-based deduction schemes in many expert system shells. Use of EXPROD reduces the expertise required for problem diagnosis but does not eliminate the need for some expertise. The purposes of this paper are to illustrate the use of expert system technology to diagnose problems with rod-pumped wells, to describe the mechanism used for problem solving, and to show how EXPROD provides trouble-shooting recommendations. Introduction Rod pumping is the most common form of artificial lift in the world. In the U.S., more than 85% of all producing wells are on rod pump. The need to identify problems quickly and accurately is essential in attempts to minimize operating costs and to maximize production. This need was the primary motivation for developing EXPROD, a tool that allows a high level of expert knowledge to be applied daily in the field. EXPROD, a medium used to record expertise permanently, is improved continuously by addition of new knowledge as verified field cases are made available. EXPROD was developed as a companion to the surface and downhole analysis program, SADA. SADA. SADA calculates surface equipment loading, including gear-reducer torque analysis, motor-size analysis, unit balancing, rod stress loading. and a downhole dynamometer card based on a measured surface dynamometer card. Other parameters are also calculated; a complete SADA printout is shown in Figs. 1A through 1C. Because the wave-equation solution used in SADA does not fully account for friction between rods and tubing, the load range of the downhole pump dynamometer card is expanded beyond the true fluid load on the plunger, and the net fluid producing stroke may be shorter than the gross pump stroke. To obtain the true fluid load on the pump plunger and the net pump stroke, the user must apply judgment to set horizontal and vertical lines interactively on the downhole pump card during a SADA program run. Important parameters, such as pump intake pressure and fluid level as shown in Fig. 1B, are derived from these line settings. To diagnose problems accurately and to suggest corrective action with SADA alone, the user must have considerable rod-pumping expertise. Because many SADA users have little rod-pumping expertise, the diagnosis step often is a problem. Application of EXPROD. EXPROD was developed to make rodpumping expertise available to field-operations personnel. EXPROD, in combination with SADA, is a significant improvement over SADA alone, but it is not a cure for all problems. Inaccurate data can result in a wrong diagnosis. To obtain the maximum benefit from EXPROD, the user must he aware of such problems as inaccurate production gauging and other input data problems. The user also needs to he familiar with a well's past production history and local operating practices to judge the validity of EXPROD's diagnosis. EXPROD's diagnostics complements the skilled user but is not a replacement for a real expert. Computing and Data Requirements Computing Requirements. EXPROD was written in BASIC for the HP 9816 microcomputer. The choice of language and computer system were made to ensure compatibility with the previously developed SADA program and with about 30 dedicated microcomputers used by field personnel for rod-pumping diagnostics in the field. Most of the HP microcomputer systems we use have about 500 K of available memory. Data Requirements. Because of the memory limitation of the HP microcomputer used, EXPROD could not he fully integrated into the BASIC program code of SADA. To use EXPROD, the SADA program is run first and the analysis results are stored on disk in a well data file. EXPROD reads this file and performs diagnostics for either one well at a time or, in batch mode, several wells per run. In addition to the data shown in Figs. 1A through 1C, the following information is useful for the EXPROD analysis: tubing size and whether or not the tubing is anchored; the likelihood of gas interference; net profit per barrel of oil; expected oil production when problem is solved; motor size; and service factor for rodstring stress analysis. If any of these additional data are unknown, EXPROD can still be run, but the diagnosis is often limited or several possibilities may be presented as alternative causes of a particular problem. Scope of EXPROD's Calculations EXPROD recognizes common pump-dynamometer-card shapes - such as full pump, fluid pound, gas interference, and leaking valves-through a statistical pattern recognition scheme coupled with a network of rules derived from company expertise. EXPROD considers such data as fluid production, gross pump stroke and maximum polished-rod load to help identify tubing leaks, worn pumps, and stuck pumps, types of problems not always apparent from pump-dynamometer-card shape alone. EXPROD can detect difficult-to-identify problems, such as the load cell being out of calibration when the dynamometer data were recorded, and it ensures that all detected problems are presented to the user. This is an important feature because SADA users identifying a major problem, like a leaking valve, may overlook an over-loaded rod section or an improperly sized prime mover. With EXPROD, inexperienced field personnel have more confidence when making recommendations to solve rod-pump problems. Important production parameters-e.g., pump intake pressure, fluid level, and net pump stroke-are estimated by EXPROD. These estimates were previously made by users during the interactive SADA pump-card analysis. These estimates are often the most difficult and subjective part of the entire data-analysis procedure. The ability to run both EXPROD and SADA in batch mode makes these programs well suited for remote rod-pump-system diagnostic analysis because the procedure of data collection and problem detection can be automated. JPT

科研通智能强力驱动
Strongly Powered by AbleSci AI
更新
PDF的下载单位、IP信息已删除 (2025-6-4)

科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
JFy完成签到 ,获得积分10
1秒前
zzyytt完成签到,获得积分10
3秒前
和光同尘完成签到,获得积分10
5秒前
djshao应助xdx采纳,获得10
6秒前
HHH发布了新的文献求助20
6秒前
xxw完成签到,获得积分10
7秒前
浮若安生完成签到,获得积分10
7秒前
8秒前
loren完成签到 ,获得积分10
8秒前
lit完成签到 ,获得积分10
9秒前
djshao应助222采纳,获得10
9秒前
10秒前
安宁完成签到 ,获得积分10
10秒前
10秒前
思源应助坦率的电灯胆采纳,获得10
11秒前
12秒前
13秒前
林夕完成签到 ,获得积分10
13秒前
14秒前
蓝天发布了新的文献求助10
14秒前
16秒前
whk发布了新的文献求助10
16秒前
爱喝水的乌鸦完成签到 ,获得积分10
16秒前
迷路豁完成签到 ,获得积分10
17秒前
舒心幻竹完成签到 ,获得积分10
17秒前
强健的绿蕊完成签到,获得积分10
18秒前
ZXF完成签到 ,获得积分10
19秒前
Atropine完成签到 ,获得积分10
22秒前
茗泠完成签到,获得积分10
22秒前
duang发布了新的文献求助10
23秒前
yn发布了新的文献求助10
23秒前
25秒前
小刚完成签到,获得积分10
25秒前
fantastic完成签到,获得积分10
25秒前
茗泠发布了新的文献求助30
26秒前
包容友儿完成签到,获得积分10
26秒前
牛乃唐完成签到 ,获得积分10
27秒前
愤怒的蚂蚁完成签到,获得积分10
27秒前
不去明知山完成签到 ,获得积分10
29秒前
30秒前
高分求助中
(应助此贴封号)【重要!!请各用户(尤其是新用户)详细阅读】【科研通的精品贴汇总】 10000
Encyclopedia of Agriculture and Food Systems Third Edition 2000
Clinical Microbiology Procedures Handbook, Multi-Volume, 5th Edition 临床微生物学程序手册,多卷,第5版 2000
人脑智能与人工智能 1000
King Tyrant 720
Silicon in Organic, Organometallic, and Polymer Chemistry 500
Principles of Plasma Discharges and Materials Processing, 3rd Edition 400
热门求助领域 (近24小时)
化学 材料科学 生物 医学 工程类 计算机科学 有机化学 物理 生物化学 纳米技术 复合材料 内科学 化学工程 人工智能 催化作用 遗传学 数学 基因 量子力学 物理化学
热门帖子
关注 科研通微信公众号,转发送积分 5600606
求助须知:如何正确求助?哪些是违规求助? 4686243
关于积分的说明 14842399
捐赠科研通 4677148
什么是DOI,文献DOI怎么找? 2538898
邀请新用户注册赠送积分活动 1505830
关于科研通互助平台的介绍 1471201