Figure 2.1: The figure suggests how the six layers of functionality, defined by
ISO-13374, could be envisioned, and how the layers can connect to databases
with historical data and models.
A decision support system for tactical level maintenance planning is expected to produce a a best-known Pareto-optimal set of alternative strategies for allocations of resources over a planned scenario. Allocation strategies will be evaluated on expected equipment availability, cost and confidence criteria. Confidence criteria describes the confidence in the suggested solution, depending on both aleatory and epistemic uncertainties in the models. Models describing expected equipment behavior are expected to utilize the latest monitoring information available to maximize the confidence in the results. The goal is to enable the customer to create a tactical level maintenance plan, and the confidence to act according to that plan, to achieve high equipment availability while using as few resources as possible. The resulting problem structure is broken down functionally using the standard ISO-13374.
The section starts with a breakdown of the functionality required for the realization of such a decision support system in Section 2.1.1. The data needed to provide the identified functionality is then described in Section 2.1.2. Section 2.1.3, details the components that are currently missing and that are addressed in this text.
The standard ISO-13374, Condition Monitoring and Diagnostics of Machines, defines six layers of functionality typically found in a condition monitoring system. The standard was developed by MIMOSA (Machinery Information Management Open Systems Alliance) and IEEE among others. The objective of MIMOSA was to create a widely adopted open standard, promoting exchange of information between maintenance providers and ultimately interoperability between hardware and software components used for condition-based maintenance (Thurston 2001). Figure 2.1 illustrates the six layers of functionality defined by the standard.
Decision support for the tactical decision level does not strictly concern condition-based maintenance actions directly, but rather resource allocation to enable them, the flow and refinement of information to generate decision support can be described using the same six functional layers provided by the MIMOSA standard. Table 2.1 suggests a brief description of functionality that occur on operational and tactical decision level at each functional layer.
The focus of this text is to study the requirements and possible solutions related to providing the functionality in the last cell in the tactical level column on layer six, in Table 2.1. The functionality mentioned elsewhere in the table contributes to the sought functionality; tactical level decision support. Key inputs to the tactical level maintenance decision making process is generated by the functions and activities on the operational decision level. Before discussing the breakdown of the functionality on the tactical level, the functionality on the operational level is summarized.
On the operational level, the functions described in layer two through six of the standard; “Data Manipulation”, “State Detection”, “Health Assessment”, “Prognostics” and “Decision Support”, can all take place onboard the maintained equipment unit. Sensors are used to collect the data required to monitor the state of the equipment. Operator and maintainer input is used to report performed maintenance actions and any observed anomalies. The timing of maintenance activities in relation to usage and environmental data is required to estimate the wear of different components. For example, a humid environment will affect the corrosive process of a gun barrel differently depending on whether it has been cleaned after shooting. The information collected by the onboard unit is refined through the functions outlined in each functional layer in Table 2.1, which ultimately generates suggestions and instructions for the equipment operators regarding maintenance. The reliability behavior of a piece of equipment during a planned scenario is affected by the maintenance actions suggested by the onboard unit, as well as the operators response to the maintenance suggestions. Eventually, the outcome is reported to the
backoffice. An aggregated status report giving an indication of the current system status may be available in near real-time, whereas a complete set of information including collected monitoring data, all performed maintenance actions and registered configuration events will likely only be made available on a less frequent basis.
One of the inputs to the tactical level maintenance decision making process, layer 1 of the column highlighted in blue in Table 2.1, consists of the data described in the previous paragraph; monitoring, usage, maintenance and configuration data from each piece of equipment. Logistics data, in the form of the status of the supply chain is also a significant input to the tactical level maintenance decision process, as it describes the availability of resources. Decisions on tactical level are made based on expected future usage, that is, a plan. A plan description based on the tactical goals, which also includes any constraints with regard to budget and/or resources, is the final input to the tactical level maintenance decision process.
To facilitate use of the data provided into the tactical level maintenance decision process, filtering and aggregation of detailed data related to individual pieces of equipment is done. The result of this process will generate updated information regarding the status of the equipment on the operational level, which may affect the reported status of the supply chain since spares and resources may have been consumed since the last update. These steps constitute layer 2 and 3 of the tactical level in Table 2.1.
Health assessment, layer 4, involves determining the state of the supply chain in relation to identified immediate needs and determined or contracted minimum levels of spares and resources at each location. It also involves performing the same analysis on a higher level, determining whether resources that are procurable, or in central storage, are adequate to meet the currently identified demand. The tactical level planning functionality identified in this text involves making a new health assessment based on future planned usage. Similarly for layer 5, prognostics regarding current lead-times and currently available capability and capacity will be updated with future estimations based on a planned scenario. For example, lead-times will vary depending on the planned physical location of resources and equipment.
Tactical level decision support for maintenance planning, layer six of the highlighted column in Table 2.1, is the focus of the research in this text. It will consist of a decision support system that assists the user in identifying optimal allocation strategies for a planned usage scenario. Optimization criteria include equipment availability, cost and confidence in the proposed solution. In order to allow the user to interact with the decision support system in a cooperative manner, to integrate his or her knowledge of the problem, a best-known Pareto-optimal set of identified solutions will be generated.
The context in which tactical level maintenance planning could be performed is described in Table 2.1 Figure 2.2 illustrates the here identified flow of data to and from a tactical level maintenance planning decision support system in this context. Six different units which interacts with the decision support system can be identified; “EMMS Onboard”, “Suppliers”, “LSA / Engineering”, “Manufacturing”, “Service Organization” and the “User”. “EMMS Onboard” represents the equipment and the data that can be collected from the use, environment and maintenance activities performed on the equipment. “Suppliers” can be internal or external providers of resources, typically manufacturers of different spare parts. They are expected to provide information regarding resource availability and lead-time. “LSA / Engineering” represents the logistic support analysis and engineering functions responsible for equipment development. Equipment reliability models predicting equipment behavior under different conditions are provided from this source. “Manufacturing” provides information about the initial state of each piece of equipment “as-delivered”, the information that describes the initial configuration of components which comprise an equipment unit. The “Service Organization” is an abstraction of the different maintenance facilities involved in the maintenance of the equipment . The service organization will provide information regarding the currently available supplies in different locations. The intended “User” of a decision support system for tactical level maintenance planning is either military staff in charge of the allocation of maintenance resources for a planned activity, e.g. mission planning. The planner could also be a civilian contractor working closely with the military customer to assist with the best allocation of resources for upcoming activities.
Six different types of data are represented in Figure 2.2; “Usage”, “Environment”, “Configuration Management”, “Supply”, “Models” and “Plans”. Plans consist of planned usage in a predicted environment, along with any changes and constraints to the supply chain. Plans are in other words a super-set of “Usage”, “Environment” and “Supply”. “Usage” and “Environment” can essentially be treated as one since use implies that an activity is being performed in some kind of environment, whether that environment is predicted, measured or unknown. “Usage” and “Environment” can thus be reduced to simply “Usage”. This leaves four different types of data:
Excluded from the view in Figure 2.2 is the information flowing into the different boxes at the top of the figure. From the decision support system point of view, the provided data is expected to be a product of the backoffice. The rationale behind the exclusion of this information from Figure 2.2 is to focus explicitly on the information that goes into the decision support system at the time of planning, rather than all the data that flows within the equipment maintenance monitoring system.
Using the description of the functional layers and the input data described in sections 2.1.1 and 2.1.2 respectively, it is now possible to identify a number of components, illustrated in Figure 2.3, that are required for the realization of a decision support system for tactical level maintenance planning.
An equipment reliability model is required to be able to predict the expected maintenance related outcome of using the equipment according to a provided scenario. The equipment reliability model is typically developed during engineering of the system, and can thus be provided from the LSA-process at the manufacturer. However, it is important that the equipment reliability model can incorporate the latest information available regarding actual system performance.
A model describing the current status of the fleet of maintained equipment, and the resource distribution and lead-times within the maintenance organization, is also required. This is used to describe the available resources to be allocated, as well as predicting the aggregated relation between resource consumption and equipment availability, based on the expected fleet behavior under the planned usage profile. Depending on the arrangement between the maintenance support provider and the customer, specific resource distribution and expected lead-times may be provided to the decision support system by the planner rather than stored in the equivalent of a backoffice environment. It is imperative that this model is continuously updated using the latest available monitoring and configuration management information . Confidence in the reliability models, as well as expected lead-times of maintenance support delivery, will be affected by the availability of monitoring and configuration management data. The organizational model must thus include the effects of the planned information flow in the maintenance organization as well.
Finally, an optimization method is required in order to find the requested Pareto-optimal set of strategies for allocations of resources over the planned scenario. The method must be able to use the organizational and the equipment reliability models, information regarding current equipment and resource status, and the provided planned usage profile, to extract information regarding the requested criteria; expected availability, cost and confidence.
Based on the structuring of the problem of generating tactical level decision support provided in this section (2.1), requirements for models to describe fleet status, resource availability, effects of information flow and expected equipment behavior under a planned usage profile can be identified. These aspects are covered by the two models proposed above. Each corresponds to a separate decision level in the structuring of the problem (tactical versus operational), and so are unlikely to be suitable for description in a single model. At the same time, no other decision levels have been identified, and so the described model requirements are here argued to be minimal and sufficient for solving the problem. A similar argument can be applied to the identified requirement for an optimization process; a single problem is posed by the provided status information and the planned usage scenario, in the form of a request for a Pareto-optimal set of resource allocation strategies. A requirement for a single method to solve a single problem is minimal and sufficient.