A spacecraft can be defined as an amalgam of complex systems composed of active components that can be either mechanical, electrical, electronic or electromechanical (EEE), propulsive elements, or even passive parts [1]. Provided that any existing component is susceptible to failure at some point of their lifespan, it is crucial to conduct reliability assessments in order to ensure that the main functionalities of the spacecraft will be operational. Otherwise, the integrity of the spacecraft and the overall success of the mission would be put at risk.

Making reliability predictions takes on greater importance nowadays, given the abrupt change the space field is experiencing, driven by technology miniaturization, faster manufacturing of space products and sustainability concerns. A new horizon with more competitiveness within the space industry and open to new business ideas and innovative projects, is the ideal setting to show the importance of reliability analyzes of space systems. For instance, satellite constellations are revolutionizing the telecommunications industry, by providing global coverage and connectivity, as well as the Earth Observation field, by enabling environmental phenomena monitoring through multiple satellite usage.

In this new paradigm led by private entities and organizations in contrast to traditional government-funded space missions, there is an enhanced leaning towards Commercial-Off-The-Shelf (COTS) technologies and components usage, investing in a model focused on democratization of space by minimizing costs and reducing the duration of the development schedule of the missions.

Therefore, it is crucial to evaluate the behavior of the space systems and technologies chosen for the development of any space mission since depending on the type of failure that a system may suffer, it may affect other systems or even compromise the integrity and success of the mission, potentially endangering other neighboring spacecraft by becoming space debris.

Now that the importance of reliability predictions in space applications has been introduced, one may wonder: how can these predictions be used?

What are the primary applications of reliability prediction in space thechnology?

Reliability prediction is a type of the Product and Quality Assurance tool used in space projects to demonstrate compliance with the requirements set regarding the probability of failure of a system at a particular mission phase, as illustrated in Figure 1.

These requirements may be fixed by different entities or organizations, for instance, the European Space Agency (ESA) via the ECSS Standards, or the launching company, among others. Whether a space system is required to prove compliance via reliability analyses, testing and/or simulation depends primarily on the external stakeholders involved within the space project.

Furthermore, reliability prediction analyses are also performed at an internal level to identify design weaknesses -i.e. non-existent redundancy, and to compare different design configurations. Likewise, by assessing the failure rate of a component, it is possible to estimate the frequency of maintenance, that being repair or substitution, to guarantee a successful performance. However, this type of analyses should not be used as stand-alone assessments but should be complemented by other reliability tools such as Failure Modes Effects Analyses (FMEAs), among others, as not all the components’ characteristics are considered in one specific analysis.

Figure 1: Reliability Prediction Main Uses [2]

Reliability Prediction Analysis

RPA is not a compulsory analysis for any of the stages of the life cycle of a product, but rather a resource used when the reliability information is needed and not available. This may happen during various stages of a project such as PDR, CDR, TRR, or TRAR. Depending on the available information, RPAs can be either a ballpark estimation or a quite accurate one as they highly depend on the available data. This makes it not strange to update the same RPA as the item/equipment is more concretely defined during the progression of the project.

How is an RPA performed?

The RPA is a bottom-up method, where the item/equipment’s failure rate is computed from the combination of failure rates of the individual components. The most broadly used methodology is the Series Model, where all the components are assumed to be in series and the failure rate of the item/equipment is simply the sum of the failure rates of all the components, as defined in section 6.4.4.2 of the MIL-HDBK-338B [3]. This model provides a conservative reliability prediction as it considers that any failure of any component directly causes the item/equipment to fail.

Types of RPA

The most important step when performing an RPA is the selection of the failure rate calculation methodology. An RPA is done when there is no empirical data on the reliability of the item/equipment, so what we are doing is trying to predict it. To do this task, it is obvious that the more information from the component available, the more accurate will the prediction be. From this can be deduced that the methodologies which require a higher number of inputs are considered the most accurate ones.

Figure 2: Reliability Prediction Analysis Methodologies

• Historical Prediction: The necessary inputs are the type of component and the environment. With these two inputs, the engineer must look in the corresponding database for the most similar component to get the failure rate. This prediction is used when the information about the component is very vague.
• Parts Count Prediction: The necessary inputs are the type of component and the environment. They are used in a prediction standard in combination with conservative technical specifications and operating conditions (both given by the prediction standard) to predict the failure rate. This prediction can be used when the information available is enough to use the selected standard.
• Parts Stress Prediction: All the input types are necessary for this analysis. All of them are inputted into the corresponding prediction standard and the failure rate is computed accordingly. Different standards may require different information.

Standard and Data book predictions

There are several standards or handbooks for reliability prediction for each type of element, that being for mechanical, electronic, or miscellaneous parts. Some of these standards and data books are depicted in Figure 3 below.

Figure 3: Reliability Prediction Standards and Data books for EEE and Mechanical Parts

FIDES

FIDES, which stands for Failure In Time/Duration Equivalent Standard, is a reliability prediction standard used in the field of electronic and electrical systems. It is often employed to estimate the reliability of components or systems in industries such as aerospace, automotive, and telecommunications.

FIDES is particularly associated with the prediction of the failure rates of electronic components. The standard provides guidelines for evaluating the reliability of components by considering factors such as temperature, stress levels, and environmental conditions. The goal is to estimate the expected number of failures over a specified period.

Now, given that that there are different reliability prediction standards as depicted in Figure 3, one may inquire: Why FIDES? What features does FIDES have to offer?

Why FIDES?

As depicted in Figure 4, FIDES approach is based on three main pillars: component technology, use and process.

Figure 4: FIDES Approach [1]

FIDES became a Standard in France in 2005, and it has since been updated twice from its initial release, as evidenced by the existence of FIDES2009 and its most recent version, FIDES2022, which was published at the end of 2023. It provides a detailed and up-to-date classification of electric and electronic components for commercial and military fields, whereas its counterpart’s (MIL-HDBK-217F) latest update was in 1995 [4].

Given the rapid pace of the technological advancements over the past decade, it is worth noting the fact that any handbook that is not updated periodically may soon become obsolete.

Regarding the use, the FIDES standard considers various parameters, including the environment in which the electronic components operate, the stress levels they experience, and the expected duration of operation. For instance, it considers whether a component is activated or not, and the temperature, humidity, and salinity of the ambient in every mission phase.

Likewise, considering the process followed to design and manufacture the product is paramount for reliability prediction, FIDES incorporates an audit, a list of questions to be filled out that provide an overall overview of the industrial process, in the prediction analysis.

Provided that all previously mentioned parameters significantly affect the performance of electronic devices, it is essential to consider them when developing a reliability assessment. Hence, FIDES utilizes mathematical models to make predictions about the reliability of these components. So, how is the FIDES methodology implemented?

How does FIDES work?

FIDES uses the same formula for all components composed of three different parts:

λ = λ_physical · ∏_PM · ∏_Process

• λ_physical represents the contribution from the component.
• ∏_PM takes into consideration the quality and technical control over manufacturing of the item.
• ∏_Process accounts for the quality and technical control over the development, manufacturing, and usage process of the product containing the item.

Over this general formula, λ_physical is decomposed into different terms, that depend on the component being calculated.

• Base failure rate: Depends on the general type of component, in this case, the type of resistor.
• Physical stress factors: Accounts for the stress suffered by the component caused by its operation and by environmental factors.
• Induced factor: It is computed based on the overstresses inherent to the application field.

As an example, the following formula is used for computing the failure rate of a resistor:

λ_physical = λ₀ × ∑_i^phases (t_annual / 8760)_i × (Π_{Thermo-electrical} + Π_TCy + Π_Mechanical+ Π_RH)_i × (Π_Induced)_i

FIDES 2009A

FIDES 2009A is the 2^nd release of FIDES, and considers the following components:

Electronic components

• Integrated circiuts
• Application Specific Integrated Circuit (ASIC)
• Discrete Semiconductors
• Light emitting diodes (LED)
• Optocouplers
• Resistors
• Fuses
• Ceramic capacitors
• Aluminum capacitors
• Tantalum capacitors
• Magnetic components: Inductors and Transformers
• Piezoelectric components: Oscillators and Quartz
• Monostable electromechanical relays
• Switches
• Printed circuit board (PCB)
• Connectors

Hybrids and Multi Chip Modules

• Micro-components
• Wiring, case, substrate, external connections

Microwave (HF) and radiofrequency (RF) components

• RF HF integrated circuits
• RF HF discrete semiconductors
• RF HF passive components

COTS boards

Various subassemblies

• LCD screens (TFT, STN)
• Hard disks (EIDE, SCSI)
• CRT screens
• AC/DC and DC/DC voltage converters
• Lithium and nickel batteries
• Fans
• Keyboards

A complete case study in RPA

If you are willing to follow the process of building a Reliability Prediction Analysis using the FIDES standard, we have build an example Case Study using the reliability tool Robin RAMS. The Case Study develops the reliability calculations for a generic Data Acquisition Module of a satellite, starting from a Bill of Materials.