Destroying Value in Manufacturing with Poor Decisions
I started to write the title for this post as "Top 3 value destroying behaviours...." but quickly stopped myself due to a rabid dislike of any articles stating Top..... anything. "Top, Most or any other claim should be limited to survey data results with some statistical legitimacy so I am sure you will support me in avoiding that! Rant over.
In all seriousness, performance pressures on manufacturing companies continue to increase globally and in particular in high labour cost countries such as Australia. Intensifying global competition, reduction in technical resources due to ageing population, industry consolidation, stricter environmental regulations, higher raw material costs, plummeting product life-cycles, and increasingly demanding customers and shareholders are only a few of the issues plaguing manufacturers.
These broad ranging factors represent areas of continuous improvement focus for manufacturers to ensure they survive and thrive. Innovation is occurring rapidly and the Industrial Internet of Things, device connectivity, big data, virtualisation and other trends are tackling these challenges from a range of different angles. But what about some lower hanging fruit? There is plenty of ripe opportunity in most manufacturing environments for making better decisions, particularly in smaller tier companies.
In specific areas of the production floor and the associated schedules for the operation there are behaviours and value leaks that are endemic in many manufacturing companies. Tell me if you agree or not:
Between silo'd operations
For any multi-step process entity, manufacturing supply chains are split up into silo's for geographical, regulatory, commercial or complexity reasons. These silos can exist over large (global) or short (metres) physical distance or be completely virtual. Different managers responsible for closely linked silos often operate to a different beat, and work to misaligned KPI's. Production decisions are made in isolation without understanding the impact on the performance of the overall system. Unified KPI's are often lacking and optimal performance of one silo generally destroys value in another. Decision makers are happy to "settle" for what a precursor silo tells them and sub-optimal schedules result without sufficient what-if modelling to challenge what is best for the business. A classic example is the walls put up in a process industry such as dairy. Raw material management (intake) does not sync with what's best for the processing facility, which in turn does not fully model the dynamic with warehousing, inventory and outbound logistics. This becomes more complex in a multi-plant regional setup.
These relationship dynamics obviously drive the requirement of integrated value chains though often, even with aligned systems, processes and metrics, the problem still persists. We hear time and time again that it is difficult to manage complex operations at the required level of detail, particularly considering the dynamic nature of near real time requirements across manufacturing processes.
Within manufacturing process areas
Drilling down a little deeper, between individual manufacturing processes or machines, the sequencing of orders and associated routes can quickly multiply to be an enormously complex problem. Optionallity can be mind boggling and we often see planners and schedulers resorting to their own tribal beliefs on how to schedule. This results in "status quo" embedded business processes such as "I always schedule that Product/SKU using that route because.....
Often, the tribal view may very well be the best, but if we are not continually challenging, simulating and assessing alternate operating strategies, how do we really know?
An example of this behaviour was exposed recently where a production scheduler was responsible for managing a complex 10-line environment and did a great job sequencing production orders in a multi-million dollar enterprise finite capacity scheduling software system from an ERP provider. Orders were dutifully placed on lines, constraints were managed. There was a dynamic interplay between staff - operators, procurement, maintenance, sales etc interacted via systems, verbal discussion and production meetings prior to committing the schedule. When we looked in detail, there were a couple of red flags. Resource and route capacities were set once a year based on financial budget models and there was high variation between expected run times and actual. Of course OEE metrics looked great and everybody was patting themselves on the back. Schedule adherence was almost always positively skewed (a strong leading indicator of something wrong) and changeover times were static and always within spec. Imagine though, the scheduler creates a conservative schedule but it is based on conservative modelling of actual capacities, not updated regularly enough - the operators receive this easily achievable schedule, complete each order, look at the expected changeover time and maybe, just maybe, take it a little easy between operations. Before you know it you have a cascading and accumulating inefficiency that is difficult to measure in the defined KPI's. Huge latent capacity or cost reduction can be gained through rigorous focus on actual to plan performance and regular updating of the tolerances on decision models based on real world operational data, not last years modelled financial metrics.
Departmental walls are commonplace. My favourite is the systematic lack of robust collaboration for production order and maintenance order sequencing. It is rare to see a completely integrated production scheduling and maintenance scheduling system, balancing the most efficient, cost effective and proactive maintenance strategy against the all important production impact. We hear a lot of rhetoric about carrying out maintenance at end of shift times, or waiting until the weekend to perform cleans/changeovers/ maintenance tasks. When asked, how management know if that is the best time to carry out this work, the answer invariably is they don't and it was a call made by the business to run that way.
Why don't scheduling tools integrate seamlessly with maintenance systems? System Integration technologies are vastly superior and more efficient than even 5 years ago. Is it too much to ask that the production scheduling system has real time visibility into the current planned or required work against each resource? And that the production scheduler and maintenance planner can run multiple what-ifs to determine "best time" outcomes? Phone calls, distributed spreadsheet reports and arbitrary judgement calls are still commonplace and they do not need to be.
The Way Forward
We see investment in a few key areas as critical for enabling quick operational improvement without jumping on the latest trend or cutting edge technology. Understanding manufacturing plant performance, measuring downtime and modelling appropriate OEE metrics and implementing processes to improve these is critical. But importantly, not doing it in a way that can be easily influenced by tribal behaviour and human influenced adjustment - so automated systems that remove human error are necessary. Then using this information to continuously validate decision making systems such as Master Production Scheduling, Detailed Scheduling and Execution tools such that the models are more reflective of the shop floor. The barriers to entry for these capabilities have traditionally been high (read expensive with significant ongoing management), but innovative and disruptive technology vendors are rapidly reducing the price point and ease of integration, taking away market share from expensive providers and systems integrators who thrive on long, complex and drawn out projects for deriving revenue.