01-08-2013, 12:47 PM
Sensor Network Based Oilwell Health Monitoring and Intelligent Control
Sensor Network Based Oilwell .pdf (Size: 820.94 KB / Downloads: 80)
Abstract
Most oil pumping units (OPUs) have been using
manual control in the oilfield. This existing oil-pumping system,
a high power-consuming process, has the incapability of OPU’s
structural health monitoring. In this paper, a sensor network
based intelligent control is proposed for power economy and
efficient oilwell health monitoring. The proposed sensor network
consists of three-level sensors: 1) several types of basic sensors,
such as load sensor, angular sensor, voltage sensor, current sensor
and oil pressure sensor, which are the first level sensors (FLS), are
used for oilwell data sensing; 2) our developed intelligent sensors
(IS), which belong to the second level sensor, are designed mainly
for an oilwell’s data elementary processing, main fault alarm/indi-
cation, typical data storage/indication, data/status transmission up
to the third level sensor (TLS), data/status transmission between
IS, and command transmission down to the OPU motor; and
3) our developed software-defined (SD) control centers with an
embedded database, i.e., the TLS, are designed for hundreds of
oilwells data storage/management, data processing, malfunction
detection, malfunction alarm/indication, stroke-adjustment com-
mand transmission down to a specific IS for power economy and
the malfunction report to the maintenance staff via global system
for mobile communications (GSM) short message service (SMS).
Experiment results at the Chinese Petroleum’s Changqing Oilfield
demonstrate our proposed sensor network based system.
INTRODUCTION
N the last few years, sensor networks have drawn much
attention for their broad practical applications [1]–[19].
[1]–[3] investigate specific sensors and sensor networks for air-
craft structural health and performance monitoring. A highway
bridge assessment using an adaptive real-time wireless sensor
network is proposed in [4]. A real-time radiological area
monitoring sensor network is developed in [5] for emergency
response. A congestion-aware, loss-resilient biomonitoring
sensor network is reported in [6] for mobile health applications
in the medical field. A novel interstitial fluid sensor network is
provided in [7], [8] for remote continuous alcohol monitoring.
A spatiotemporal field monitoring application prototyping
framework is given in [9]. [10] uses an energy-efficient wireless
sensor network for localization based structural health mon-
itoring. A low-cost, low-data-rate rapid structural assessment
network is shown in [11] for civil structural health monitoring.
[12] validates the horizontal dilution of precision (HDOP) mea-
sure for impact detection in sensor network-based structural
health monitoring.
NETWORK TOPOLOGY AND SYSTEM DESCRIPTION
Fig. 1 shows the proposed network topology for oilwell health
monitoring and OPU intelligent control. The proposed system
is comprised of our developed SD TLS, each of which wire-
lessly communicates with hundreds of IS. Each IS is designed
with the capability of data transferability with a set of FLS, its
adjacent IS and its corresponding TLS as well as the capability
of command transmission down to its OPU motor. Each group
of FLS, including a load sensor, an angular sensor, a voltage
sensor, a current sensor and an oil pressure sensor, are utilized
for data sensing from an OPU, which convert all measurements
into electrical signals and then transport them into its corre-
sponding IS.
DEVELOPMENT OF IS
A. System Description of IS
This subsection aims at clarifying the logic connection be-
tween the shown oilwell in Fig. 1 and the developed IS. As
shown in Fig. 2, a group of FLS in our proposed system consists
of a load sensor, an angular sensor, a voltage sensor, a current
sensor and an oil pressure sensor while the IS mainly contains
two components: the designed control board and the frequency
converter. Five kinds of sensing data from FLS are imported to
its IS. The IS usually transmits oilwell static parameters (At the
initial stage), significant malfunction reports (if necessary), dy-
namic sensing data and elementary processing data directly to
the TLS. As a special case, when the wireless communication
between the IS and the TLS fails, the IS sends data to its adja-
cent IS for relay transmission up to the TLS.
System Description of TLS
This subsection aims at clarifying the connection between the
TLS and its radio station, i.e., GD230V-8, as well as its GSM
module, i.e., GSM Wireless Module TC35i [38]. As shown in
Fig. 8, the TLS includes three components: 1) a user interface
for interaction; 2) some embedded algorithms for wireless com-
munication between the TLS and the IS, a regular data request
on all managed IS, a malfunction diagnosis, a pumping stroke
adjustment and GSM SMS; and 3) a database for data storage.
The wireless data, usually including dynamic sensing data
and significant malfunction reports for one specific OPU, is ac-
quired via the communication protocol and is then stored in its
database. The TLS is developed using Borland Delphi [30], [31]
and its database is designed using Microsoft SQL server 2005
[32], [33]. After a thorough malfunction detection in the TLS,
once a malfunction is identified, it is immediately sent to the
maintenance staff via a GSM AT command [34], which gener-
ates a corresponding short message transmission. Furthermore,
after a thorough data processing,
Malfunction Diagnosis
This project considers the 9 most important oilwell malfunc-
tions [35], [36], including (1) underground oil shortage, (2) gas
effect, (3) oil pump on the touch, (4) oil pump under the touch,
(5) wax deposition, (6) pumping rod broken off, (7) plunger
stuck, (8) oil pump serious leakage and (9) no malfunction.
Different malfunctions have quite different LPD. The typical
LPD of these 9 malfunctions are illustrated in Fig. 12, where
no malfunction corresponds to a quasi-parallelogram LPD; un-
derground oil shortage corresponds to a ‘gun’-shape LPD; a
circular arc occurs in its LPD under a gas effect; a top-right
longhorn occurs for an oil pump on the touch while a lower-left
longhorn occurs for an oil pump under the touch; some irreg-
ular dog teeth occur for wax deposition; the quasi-parallelogram
LPD becomes much narrower if the pumping rod breaks off; the
LPD becomes a cucumber-shaped tilt if the plunger is stuck; and
the top-right corner is gone under the condition of oil pump se-
rious leakage. Evidently, the oilwell malfunction diagnosis can
be executed based on LPD classification.
CONCLUSIONS
In this paper, a sensor network based oilwell remote health
monitoring and intelligent control system was proposed for
OPU management in the oilfield. This proposed system con-
sists of three-level sensors: the FLS, the IS and the TLS. The
FLS have been used for an oilwell’s data sensing, including
a load sensor, an angular sensor, a voltage sensor, a current
sensor and an oil pressure sensor for each oilwell. The IS was
designed mainly for an oilwell’s data elementary processing,
main fault alarm/indication, typical data storage/indication,
data/status transmission up to the TLS, data/status transmission
between IS, command transmission down to the OPU motor.
And the SD TLS was designed for hundreds of oilwells’ data
storage/management, data processing malfunction detection,
malfunction alarm/indication,