Abstract— The critical importance of sustaining fault diagnosis, as a major system tool, is unquestionable if the high performance and reliability of increasingly-complex engineering systems is to be sustained over time and across a wide operating range. However, it is quiet difficult to retain the joint ability of fault detection and isolation as it requires a strong system architecture. This is why, before designing an industrial supervision system, the determination of a system’s monitoring ability based on technical specifications is important as finding the source of the failure is not trivial in systems with large number of components and complex component relationships. This paper presents an efficient and cost-effective fault detection and isolation (FDI) scheme that evolved from an earlier one [23]. FDI specifications are translated into constraints of the optimization problem considering that the whole set of Analytical Redundancy Relations has been generated, under the assumption that all candidate sensors are installed and later on tested by an optimization algorithm using of linear programming and relaxed versions of non-linear programming. By doing so, the critical information about the presence or absence of a fault is gained in the shortest possible time, with not only confirmation of the findings but also an accurate unfolding-in- time of the finer details of the fault, thus completing the overall diagnostic quality monitoring picture of the system under test. The proposed scheme is evaluated extensively on a two-tank process used in industry exemplified by a benchmarked laboratory scale coupled-tank system. Keywords—Sensor Location, Optimization, Fault Detection, Isolation, Analytical Redundancy Relations, linear programming, Benchmarked laboratory-scaled two-tank system. I. INTRODUCTION rocess faults, if undetected, have a serious impact on process economy, product quality, safety, productivity and pollution level. In order to detect, diagnose and correct these abnormal process behaviors, efficient and advanced automated diagnostic systems are of great importance to modern industries. Fault diagnosis and process supervision are an increasingly important topic in many industrial applications and also in an active academic research area. Considerable research has gone into the development of such diagnostic systems [1]. Most approaches for fault detection and isolation (FDI) in some sense involve the comparison of the observed behavior of the process to a reference model. The process behavior is inferred using sensors measuring the important variables in the process. Hence, the efficiency of the diagnostic approach depends critically on the location of sensors used for monitoring process variables. The emphasis of most of the work on model-based fault diagnosis has been more on procedures to perform diagnosis given a set of sensors and less on the actual location of sensors for efficient identification of faults. The problem of sensor placement for FDI consists of determining the optimal set of instruments such that a predefined set of faults are detected and isolated. In many cases, this set is defined in order to design some remedial actions such that the control loop is able to continue operating even in the presence of a fault (fault-tolerant control). II. RELATED WORKS In sensor location optimization, the usual objective to minimize in the sensor placement problem, is the sensor cost. There are several articles devoted to the study of the design of sensor networks using goals corresponding to normal monitoring operation. Aside from cost, different other objective functions such as precision, reliability, or simply observability were used. Different techniques were also used, such as graph theory, mathematical programming, genetic algorithms and multi-objective optimization, among others. The problem has also been extended to incorporate upgrade considerations and maintenance costs. In [2], it is being noticed that the problem of sensor placement in the model-based FDI community is still an open problem. However, some contributions have already been done in this direction [3], [4], [5], [6], [7], [8], [9], among others. In model-based Fault Detection and Isolation (FDI), faults are modeled as deviations of parameter values or unknown signals, and diagnostic models are, in such cases, often brought back to a residual form. For works based on continuous differential/difference-equation- based models (see, e.g., see [1] and [10] and the references therein for discrete-event models [11], [12] and for diagnosis of hybrid systems [13]. To be able to perform model-based supervision, some redundancy is needed, and this redundancy is typically provided by Sensor Location Optimization for Fault Diagnosis with a Comparison to Linear Programming Approaches R. Doraiswami, Lahouari Cheded, Haris M. Khalid, M. Akram, Hassini El-Kafi and Amar Khoukhi P DISCLAIMER: This is the authors' version of an article published in early access of International Conference on Computer, Electrical and Systems Engineering (ICEME) as CONFERENCE PAPER. Changes were made to this version by the publisher prior to publication.