УДК 620.179.16
PROBABILITY OF DETECTION SIMULATIONS TO STUDY THE
INFLUENCE OF SURFACE ROUGHNESS ON THE RELIABILITY OF ULTRASONIC TESTING SYSTEM
M.S. Mohammed*, Kim Ki-Seong** Department of Mechanical Design Engineering, Chonnam National University,
Yeosu 550-749, Korea * E-mail: msiddeq@gmail.com ** E-mail: sngkim@chonnam.ac.kr, Corresponding Author
Abstract. The influence of the material's surface roughness on ultrasonic echo signals has been well studied and documented in the literature. However, these studies were mostly based on the monitoring of absolute value of the echo amplitude, which cannot quantify a defect's detect-ability. In this work, the influence of the surface roughness on the Probability of Detection (POD) of a defect is studied using POD simulations to provide quantitative and insightful analysis. Furthermore, POD simulations are used to quantify the influence of surface roughness on the parameters of ultrasonic testing procedure. Steel blocks having non-metallic inclusions type of defects was inspected by simulations using CIVA software. The inspection used the ultrasonic immersion pulse-echo setup with straight beam and angle beam techniques. Comprehensive POD curves are obtained, discussed and analyzed.
Key words: POD simulation, ultrasonic testing, surface roughness, inspection qualification.
РАСЧЕТЫ ВЕРОЯТНОСТИ ДЕТЕКТИРОВАНИЯ С ЦЕЛЬЮ
ИЗУЧЕНИЯ ВЛИЯНИЯ ШЕРОХОВАТОСТИ ПОВЕРХНОСТИ
НА НАДЕЖНОСТЬ УЛЬТРАЗВУКОВЫХ СИСТЕМ КОНТРОЛЯ
М. С. Мохаммед, Ким Ки-Сеонг
Отдел разработки механических систем, Национальный университет
Хоннам, Корея
Влияние шероховатости поверхности исследуемого материала на у.з. эхосигналы хорошо изучен, однако описанные в литературе исследования опираются на изучение абсолютного значения амплитуды эхосигнала, которое не всегда определяет выявляемость дефекта. В работе изучено влияние шероховатости поверхности на выявляемость дефекта (POD) с использованием количественного анализа сигналов. Анализу с применением программного продукта CIVA подвергнуты результаты контроля стальных блоков с дефектами, заполненными неметаллической средой. Применена иммерсионная схема контроля с нормальным и наклонным падением пучка. Получены и проанализированы многочисленные кривые POD.
Ключевые слова: расчеты POD, у. з. контроль, шероховатость поверхности.
1. INTRODUCTION
In spite of the high relevance and value of Probability of Detection (POD) curves in the reliability assessment of Non-Destructive testing (NDT) systems, they have limited use. The POD is not a new methodology; nevertheless, most NDT practitioners are virtually unfamiliar with its implementation and applications. POD curves are commonly used only in few industries in some countries, but recently, numerous publications [1—5] on them have been emerged, indicating the increasing recognition of POD methodology. More research and application are needed to highlight POD concept and methodology and to familiarize the NDT community to them.
The generation of POD curves requires large resources in terms of test specimens with calibrated defects and inspectors, making POD study an expensive process. Computer simulations can help to overcome the cost and time constraints, if provided with accurate models that can be used to perform a large number of inspections on less budget and in less time [6—8], to considerably
reduce the experiments, requiring them only for verification. Moreover, standalone simulations can be used in the technical justification [9] part of the inspection qualification where mathematical modeling is acceptable.
In ultrasonic testing (UT), the versatility of POD simulations allows a large number of specimens to be studied with a large number of influential parameters and various UT procedures, unlike the experimental POD and the conventional UT measurements.
POD simulations are used in this paper with twofold objectives.
1. Studying the influence of surface roughness on the probability of detection in ultrasonic testing.
Front surface roughness has been known to generate scattering effects, which leads to attenuation of an ultrasonic signal [10—12]. The roughness-induced attenuation has been studied for different materials, at different roughness values and with different inspection techniques [13—15]. However, the principal investigation methodologies in these studies used a limited number of test pieces to monitor the effect of surface roughness on the reflected echo amplitude. This approach can reveal the underlying physical aspects of the general phenomena, but cannot properly relate surface roughness to the detectability of defects. A POD simulation can relate the detectable sizes of a defect to a wide range of surface roughness values applied to test item according to a reliable statistical distribution.
In this paper, the probability of detection of a non-metallic inclusion is analyzed using the CIVA simulation software. Non-metallic inclusions are defects that are common in steel products; they are potentially dangerous because of their effect on the mechanical properties of the material [16], such as fatigue properties [17] and the propagation of corrosive damage [18]. Ultrasonic testing is a reliable method for detecting non-metallic inclusions, although its reliability still needs to be quantified. The POD curves became an accepted tool for quantifying the reliability ofNDT systems.
2. Analyzing the influence of surface roughness on the parameters of ultrasonic testing procedure.
The qualification of a specific NDT inspection is a systematic process carried out to ensure that an NDT system, comprises of a procedure, hardware, software and personnel, is capable of detecting and characterizing specified defects in specified test items. The inspection qualification is necessary when new techniques are applied, when the inspection environment may not meet the requirements of the reference code, and when the application of NDT is critical for facilities and human safety.
One of the sound and recognized schemes for inspection qualification is the one developed by the European Network for Inspection Qualification (ENIQ) [19], which qualifies the NDT system as a combination of practical assessment and technical justification. These guidelines were originally developed for the in-service inspection of nuclear plants, and since the in-service inspection of nuclear plants demands the highest reliability, the guidelines may also be used for the qualification ofNDT of non-nuclear components or manufacturing inspections.
Technical justification is a very important part of the qualification process, which compensates the limits of the practical assessment. It provides a set of evidences about the reliability of an NDT system, evidences including computer modeling. However, technical justification is a qualitative approach that mainly relies on the expertise of the experts involved in the process. One of the computer modeling options is POD simulation, which can provide substantial value to technical justification in terms of analyzing the parameters that influence the system to determine the essential ones [20].
The second objective of this study is to use POD simulations to quantify the analysis of some parameters of UT systems (wave mode and probe frequencies)
5 Дефектоскопия, № 4, 2014
under uncertain surface roughness values, to boost the confidence of the inspection qualification process. Furthermore, it is expected that the POD simulation approach will provide useful knowledge to NDT engineers and inspectors who are unfamiliar with POD concepts.
2. PROBABILITY OF DETECTION CURVES AND THEIR SIMULATION
A POD curve quantifies the reliability of an NDT system by providing a graphical relationship between the probability of detection and the factors that control it. The detectability of a defect is related to the defect size, defect geometrical characteristic or many other physical and operational conditions [21, 22]. The POD curve provides information in the form of a probability estimated at a specified confidence bound; for example, Fig. 1 explains that for a given range of defect sizes, a defect of 1 mm length will have a 90 % probability of detection with a specified confidence (90 or 95 %) at specific values of associated uncertain parameters, e.g., defect orientation, defect dimensions, defect roughness and surface roughness. The quantitative and graphical form of the result makes it easy for NDT practitioners to understand and interpret the results. This simplicity promotes the use POD curves.
Fig. 1. POD curve example.
The results of an inspection may be recorded in two different formats: hit/miss format (i. e., the defect is detected or not detected) and signal response format (i. e., the signal amplitude is recorded to give more information about the defect size).
In the case of the hit/miss format, the POD is estimated by calculating the ratio of the detected defects to the total number of the inspected defects for each defect size. Then, the POD curve is estimated from an assumed function. The function that best fits the hit/miss data was found to be the log-odds function, so the POD model can be written as [23]:
POD(a) = <¡1 + exp
in ^
n I ln a - ^
43
where a is the defect size, ^ is the natural algorithm of the defect size for which there is a 50 % probability of detection, and c is a scale parameter that determines the flatness of the POD function.
For the signal response POD, it is assumed that the logarithm of the signal amplitude is linearly correlated with the logarithm of the detected defect size. The random error between the predicted and measured signal amplitude is supposed to have normal distribution whose standard deviation does not depend on the defect
size. These assumptions show that the POD curve can be modeled by a cumulative log-normal distribution function, which can be written as [23]:
POD(a) = 0[(lna - ^)/c)],
where O is the standard normal cu
Для дальнейшего прочтения статьи необходимо приобрести полный текст. Статьи высылаются в формате PDF на указанную при оплате почту. Время доставки составляет менее 10 минут. Стоимость одной статьи — 150 рублей.