Analytics for process design and improvement

https://www.pexels.com/search/analytics/
ISE Magazine Volume : 49 Number: 06
By Timothy Stansfield, Ronda Massey and Andrew Aitken

A systematic approach to operational intelligence

Our goal is to provide a specific approach to analytic intelligence that is realistic, results-oriented and cost-effective. A fundamental requirement of success is a proven systematic approach, a developed template that any organization can implement to meet its competitive challenges. The measures of success include the outcomes of the long-term intelligence improvements as well as the cost and timing of the long-term processes to get there.

1. Pick the members and champion of your performance/ operational intelligence team. The selection of a champion and team members, not to mention leadership and technical support for this type of assignment, often is driven by individual business cultural processes. Three specific recommendations could help ensure the long-term success of this endeavor. The first is to select a team leader who believes in statistical modeling and has a practical understanding of complex statistical modeling’s implications. Second, select a technical member or partner who has a proven record of analytics success in production processes. Often, this is not a current employee. Since this person’s role might be project-specific, it’s less important to have a full-time worker on board. Third, establish a pilot team and plan to roll the outcomes into the current management’s standard operating procedures. The organization can establish new pilot teams as programs roll to the other assets targeted for improvement.
2. Identify the project objectives and expected outcomes of the modeling effort. The selected team must determine specific and challenging outcomes for the assignment. The intent is to ensure that the modeling outcomes are realistic, relevant, cost-effective and understandable. This is often the perfect time for the team to decide whether the modeling efforts will cost too much, if this is the appropriate time or if the needed analytical data is generally available. Depending on the project’s objectives, a simplified design assessment may be an appropriate route.
3. Map the intended analytical simulation process in detail. A process mapping exercise can help the team establish a clearly defined set of goals, along with creating a unified understanding of the system. This document must include routine and nonroutine process steps, queueing requirements and options, cycle times and variances, statistical models of interference times and occurrences, labor requirements and limitations, and appropriate production inputs such as volumes, mix and variations. The process of mapping the intended situation will challenge the seemingly obvious inputs and parameters, as well as establish a priority for what analytics are needed to make sure the modeling solutions have the desired effects.
4. Prioritize the analytics selected for the simulation models. Using historical analytics can remove the opinions and assumptions of the team, helping move the discussion to a strictly data-driven model. The sources of data analytics can be similar processes with production monitoring, historical PLC data of capital performance, inventory status and quantities from ERP systems, historical staffing records, etc. These analytics can be collected across minute time frames and clearly define the statistical performance implications of labor, capital, inventory, model mix, etc.
5. Select the appropriate simulation technology and partners. The unique aspects of this type of process design assessment often require teaming partners with investment in technology. The simulation modeling and analytics consulting options are often driven by the marketing programs of these technology organizations. Selecting the appropriate partners is critical to cost-effective designs, timely and relevant decisions and long-term success. Enterprises should base these selections on the partners’ cultural fit with the organization, situational production and process experience, and their proven technical capability.
6. Develop the operational intelligence simulation and assess specific costs and benefits. The model development process will require individual team member expertise, and effectively communicating the programming, analytics, output reporting and statistical transparency can help develop trust. The modeling statistician should provide the team with a timely summary of the analytic sources, statistical summaries and explanations and required inferences. There’s always the risk that the team could see simulation modeling as nothing more than complex programming and code, so modeling experts should constantly refer to and revise the process mapping to ensure a cohesive team effort.
7. Redefine and revise the operational intelligence simulation models to ensure long-term relevance and success. The design efforts of this team are based on the best data and collective knowledge members can collect. Plan and expect timely gates of significant team decisions. However, details and revisions should be refined and revised throughout the assignment. The ongoing predicted operational costs can begin with an optimal solution that can be improved as additional analytics and models are recycled and adapted by new analytics.
8. Develop and document the long-term roles, responsibilities and measures of analytics success. Organizations should use analytics and simulation modeling in a selective manner. This tool can be improved for future use in terms of proper assignment selection, individual capabilities and passions, consistent statistical and simulation models and processes, standard operating procedures and defined team expectations. Carefully measuring lessons learned and refining this process will ensure the organization’s modeling success. In addition, an organization’s experience with these design assessment tools will determine appropriate timing and milestone expectations during future assignments.