A Unique Predictive Modeling Development Platform

The unique SPAR platform is an invaluable modeling tool for predicting and analyzing the life-cycle of systems. Understanding system life-cycle characteristics and future behavior in advance allows assessment of the cost-effectiveness of utilization, logistic support, and engineering improvement scenarios before they are implemented.

	Unique SPAR Functionality In a Single Model, SPAR captures all of the following: Design of the System Reliability of the Components Operational scenarios Logistics Infrastructure Spares Strategy Maintenance Policies … and the complex interactions between them Extensions to Monte Carlo Technology Asset Op Tempo and performance Component aging and full or partial rejuvenation by repair Custom Logic in response to events Insightful Outputs Availability/Reliability Sensitivity - Contributor to Lost Performance User-Defined Outputs

Uniquely Comprehensive

SPAR enables realistic modeling of the intricacies inherent in industrial systems. SPAR provides a broad range of capabilities to describe system component attributes, their interactions with each other, and the logistics that support them. SPAR uses this information to simulate the future system behavior.

No one can guarantee a model’s prediction, but SPAR can increase the confidence level of predictions by keeping models as close to reality as possible. A modeling tool that restricts ability to describe a system can distort the understanding of the system’s future behavior. SPAR overcomes the limitations of analytical methods in system engineering, and its unique modeling capabilities offer the means to describe reality to the desired level of detail and ultimately ensure the most accurate results.

Unique Simulation Technique

The SPAR simulation engine is the first commercial-grade software product to use Monte Carlo (MC) probabilistic simulation techniques to optimize the life-cycle behavior of industrial systems.

Clockwork Solutions has extended MC techniques, adding proprietary algorithms that handle real-world phenomena such as component interactions, component aging, variable demands on the system, uncertain and incomplete data, and others.

The SPAR MC simulator uses a state-of-the-art unbiased random number generator to provide highly accurate results. Its sophisticated mathematical engine uses a wide range of distributions (including bathtub and non-parametric distributions), while additional age conservation procedures more accurately model real-world systems. 

Unique Data Quality Adaptation

SPAR offers several mechanisms for exploring the sensitivity of model prediction to data uncertainties. An underlying idea in SPAR is the use of a Probability Distribution Function (PDF) to describe an attribute of a component. For example, rather than stating that an LRU has a probability-of-failure of 0.8, SPAR can define its propensity for failure by a Weibull distribution with Etta=5,500 and Beta=1.73. Using a Weibull PDF allows you to describe real-world phenomena such as aging.

SPAR supports a wide range of PDF types that have been found over time to represent the time-dependent behavior of failure, repair, shipment, replacement and other time-dependent phenomena.

When there is insufficient data to determine the specific values of a distribution, SPAR’s uncertainty feature can be used to state that a particular parameter can take on a range of values. For example, the mean of the failure rate can be distributed uniformly or chi-square around the given mean. This can determine the sensitivity of model prediction to variability in the data. High sensitivity indicates a need for improved understanding. Low sensitivity means no need to improve data collection for a given component.

Unique Model Validation

A SPAR model is described using Reliability Block Diagrams (RBD - akin to flow diagrams) - statistical distributions and operation rules expressed in simple language. Explaining a SPAR model to someone familiar with the system being modeled is a straightforward activity. SPAR includes several capabilities for printing out a trace of its activities. By reviewing these traces, users can easily see how SPAR reaches its conclusions. Often, non-intuitive SPAR predictions become acceptable upon review of these traces.

A SPAR model computes confidence levels for its predictions, and users can trade off increased simulation time for higher model prediction confidence.

Unique Range of Scenarios

The SPAR graphical user interface allows rapid changes to the data input and the model. Each data input parameter can be changed and investigated against the calculated performances.

SPAR output metrics enable the user to identify the impact of a variety of elements on system performance. It thus enables users to understand the source of system misbehavior, and explore the value of proposed remedies. Each scenario can be archived and compared to other scenarios to investigate and achieve the best system performances.

SPAR also supports a batch capability whereby execution initiation of several models is done off-line.

Unique System Operating Rules and Logic

SPAR uses the system operating rules and logic to model the effects of events (direct effects as well as cascaded ones) on the behavior of the system in the field. Without a logic modeling capability, a model will not account for variability in the performance (and the demands for logistic resources) inherent in the behavior of practically any system.

SPAR's logic modeling capabilities include, among others, active-passive and standby relations, cannibalization of spares, induction of failures, changing the age and the load of components, and many other operations that enable illustrating any real life phenomena of the system and its supportive resources.

For example, a simple logic for airplane sorties handling is illustrated below.

Unique Output Metrics

SPAR supplies numerous predefined metrics in addition to supporting custom metrics that can be tailored to the problem at hand.

SPAR offers several graphic and tabular displays for viewing outputs from single or multiple model runs. These include availability (or throughput) versus time graphs, sensitivity charts, and sensitivity tables. Sensitivity tables quantify the contribution by LRU type to several measures of “unperformance”.

SPAR also provides a spare parts gradient table that enables the user to identify how many additional parts should be stocked, and what the expected benefits of these parts will be. It provides spare availability, waiting time for spares and other logistic related metrics.