main-content

Über dieses Buch

Supervision, condition-monitoring, fault detection, fault diagnosis and fault management play an increasing role for technical processes and vehicles in order to improve reliability, availability, maintenance and lifetime. For safety-related processes fault-tolerant systems with redundancy are required in order to reach comprehensive system integrity.

This book is a sequel of the book “Fault-Diagnosis Systems” published in 2006, where the basic methods were described. After a short introduction into fault-detection and fault-diagnosis methods the book shows how these methods can be applied for a selection of 20 real technical components and processes as examples, such as:

Electrical drives (DC, AC)

Electrical actuators

Fluidic actuators (hydraulic, pneumatic)

Centrifugal and reciprocating pumps

Pipelines (leak detection)

Industrial robots

Machine tools (main and feed drive, drilling, milling, grinding)

Heat exchangers

Also realized fault-tolerant systems for electrical drives, actuators and sensors are presented.

The book describes why and how the various signal-model-based and process-model-based methods were applied and which experimental results could be achieved. In several cases a combination of different methods was most successful.

The book is dedicated to graduate students of electrical, mechanical, chemical engineering and computer science and for engineers.

Inhaltsverzeichnis

1. Introduction

Abstract
Since about 1960 the influence of automation on the operation? and the design of industrial processes like power systems, chemical plants, and manufacturing systems has increased progressively. This development of expanding process automation was caused by an increasing demand on process performance or product quality, the independence of process operation from the presence of human operators, and the relief of operators from monotonous and heavy tasks, as well as by rising wages. The degree of automation pushed forward drastically from around 1975 when relatively reasonably priced and reliable microcomputers were available and could solve many automation problems in one device. This was paralleled by further progress in the areas of sensors, actuators, bus–communication systems, and human–machine interfaces. The improvement in the theoretical understanding of processes and automation functions also played a large role.
Rolf Isermann

2. Supervision, fault-detection and diagnosis methods – a short introduction

Abstract
The supervision of technical processes and the quality control of products is aimed at showing the present state (condition monitoring), indicating undesired or unpermitted states, and taking appropriate actions to avoid damage or accidents. The deviations from normal process behavior result from faults and errors, which can be attributed to many causes. They may result sooner or later in malfunctions or failures if no counteractions are taken. One reason for supervision and quality control is to avoid these malfunctions or failures.
Rolf Isermann

3. Fault diagnosis of electrical drives

Abstract
Electrical drives are basic components in a multitude of devices, processes, machinery and vehicles, and in the large areas of mechanical power and process engineering, manufacturing, transportation and precision mechanical devices. Their power ranges from a few mW to hundreds MW.
Rolf Isermann

4. Fault diagnosis of electrical actuators

Abstract
Actuators usually transform low-powered manipulated variables (e.g. analog voltages 0––10V, applied DC currents 0––20mA or 4––20 mA, pneumatic pressures 0:2 – –1 bar, or hydraulic pressures 0 – –150 bar) into process input variables of a much higher power level. Frequently the process input variable is a flow of energy or matter, or a force or torque. The power needed for actuating is provided by an auxiliary energy supply, which feeds the power amplifier for the actuator. The auxiliary energy can be electrical, pneumatic or hydraulic. In many cases the actuators are composed of a signal transformer, an actuator drive, an actuator transformer (gear, spindle) and an actuating device or valve, compare Figure 4.1 and [4.7]. Actuators can operate in open loop or closed loop (e.g. position or flow-control). A survey of basic structures of actuators, different types, characteristics and mathematical models is given in [4.4].
Rolf Isermann

5. Fault diagnosis of fluidic actuators

Abstract
Fluidic actuators are characterized by their rugged design and high power-to-weight ratio. They enable one to generate linear motions easily and directly by employing hydraulic cylinders and diaphragm drives. They also dissipate energy only if in dynamic operation and during holding phases generate high reaction forces with little energy consumption. Characteristics and mathematical models are treated, e.g. in [5.9], [5.13], [5.14], [5.20] and [5.21]. Methods for fault detection of two basic actuator principles, a hydraulic servo cylinder and a pneumatic diaphragm flow valve are considered in the sequel.
Rolf Isermann

6. Fault diagnosis of pumps

Abstract
Pumps are basic components in most technical processes, like in power and chemical industries, mineral and mining, manufacturing, heating, air conditioning and cooling of engines. They are mostly driven by electrical motors or by combustion engines and consume a high percentage of electrical energy. One distinguishes mainly centrifugal pumps for high deliveries with lower pressures and hydrostatic or positive displacement (reciprocating) pumps for high pressures and small deliveries. They transport pure liquids, or mixtures of liquids and solids and herewith increase the pressure to compensate, e.g. for resistance losses or enabling thermodynamic cycles.
Rolf Isermann

7. Leak detection of pipelines

Abstract
The leak detection of pipelines is a basic task for any pipeline operation. Pipelines consist usually of different sections with a length of 30–100 km and a pump or a compressor at its inlet and a tank or storage at the outlet, Figure 7.1. The available measurements are mostly pressure p 0 and p l , mass flow $$\dot{m}_0$$ and $$\dot{m}_l$$ and temperature T 0 and T l at inlet and outlet of the pipeline or a pipeline section. In some cases the sections are separated by a sliding valve with additional pressure measurements. The measured signals are transmitted to a control station by cables, optic fibres or wireless communication, sometimes as redundant lines.
Rolf Isermann

8. Fault diagnosis of industrial robots

Abstract
This chapter describes firstly how analytical symptoms can be obtained by the estimation of physical defined parameters for the six axes of a multi-axis industrial robot. Then it is shown how the analytic symptom knowledge is added by heuristic symptoms observed by the maintenance personnel as described in Section 2.3 and Figure 2.7 and how a fault diagnosis can be performed with both the analytic and heuristic symptom information by using fuzzy-logic inferencing. As industrial robots (IR) are usually servo systems with point-to-point movements or trajectory following they have sufficient dynamic excitation and therefore parameter estimation can be preferably applied for fault detection.
Rolf Isermann

9. Fault diagnosis of machine tools

Abstract
The efficiency of manufacturing systems depends to a high degree on the reliability and availability of metal-cutting machine tools. Therefore, the detection and diagnosis of incipient and abrupt faults is of high importance. Statistics with regard to failure causes show for computer-numerical-control (CNC) drilling machines that tool faults constitute 27%, CNC faults 16%, mechanical faults 5%, electrical faults 4% and others, like organizational faults 34% and lack of orders 14%, [9.10], [9.40]. Hence, tool wear, breakage and collision contribute considerably to machine tool failures. Failure statistics for turning machine tools and machinery centers around 1993 showed that CNC and electrical failures constitute about 8%, and failures in the mechanical parts, like tool carriers 25%, workpiece handling 16%. These numbers underline the importance of automatic supervision or condition monitoring of machine tools. It is not only the breakdown of the manufacturing process that counts, but also damaged workpieces and tools.
Rolf Isermann

10. Fault detection of heat exchangers

Abstract
Heat exchangers transfer heat between two or more media. They exist in a large variety of types in the chemical and power industries, in buildings and vehicles. Typical faults in these heat exchangers are leaks, e.g. by corrosion and the contamination by dirt and dissolved or suspended matter. The growth of deposits is called fouling and leads to a reduction of heat transfer. Therefore heat exchangers are usually designed with excess heat transfer surfaces of about 35% average, [10.16]. This excess design increases costs, space and weight. Remedies against fouling are chemical or mechanical mitigation techniques like filtration, additives, higher velocities, lower surface temperatures, polished surfaces. However, fouling cannot usually be avoided completely and therefore periodic cleaning will still be necessary. For more details see, e.g. [10.19], part Oc 1 and [10.17]. The detection of leaks in heat exchangers may be based on mass balances and methods described in Chapter 7. Fouling increases mainly the heat transfer coefficients, respectively the heat transfer resistance and to a minor extent the flow resistance of the media. The following sections describe some methods to detect changes of the heat transfer and some experimental results for steam-heated tubular heat exchangers with linear and parameter variable models.
Rolf Isermann

11. Fault-tolerant systems – a short introduction

Abstract
The improvement of reliability can be increased by two different approaches, perfectness or tolerance, [11.4]. Perfectness refers to the idea of avoiding faults and failures by means of an improved mechanical or electrical design. This includes the continued technical advancement of all components that increase the operational life. During operation the intactness of the component must be maintained by regular maintenance and replacement of wearing parts. Methods that facilitate fault detection at an early stage allows one to replace the regular maintenance schedule with a maintenance-on-demand scheme.
Rolf Isermann

12. Examples of fault-tolerant systems

Abstract
High-integrity systems require a comprehensive overall fault tolerance by faulttolerant components and an automatic fault management system. This means first the design and realization of redundant components which have the lowest reliability and are safety relevant. In automatically controlled systems there are, for example sensors, actuators, computers, communication (bus) systems, control and operational software and process parts, like electrical drives, tube lines, pumps or heat exchangers. Components with multiple redundancy are known for aircraft, space, train and nuclear power systems. Other technical processes with redundancy are for, example lifts (multiple ropes and brakes) or multiple pumps for steam boilers, see [12.2].
Rolf Isermann

13. Terminology in fault detection and diagnosis

Abstract
The following definitions are the result of a coordinated action within the IFAC Technical Committee SAFEPROCESS, published in [13.3]. Some basic definitions can also be found in [13.1], [13.4] and in German standards like DIN and VDI/VDERichtlinien, see references at the end of this section and [13.2].
Rolf Isermann

Backmatter

Weitere Informationen