The choke model – making the facts visible
Updated: 13 hours ago
A choke model analysis is the starting point to understanding production losses, where they originate in the process and what causes them.
In the previous article, I described short interval control (SIC)—the discipline of reviewing process performance at regular intervals during a shift and collecting information on production losses. But what are production losses and where should SIC be installed in a complex production plant with multiple processing units? In this article, I introduce the concept of choke points and explain how a choke model is used to decide where to install SIC and where to focus improvement efforts.
Lost production is defined as saleable output that could have been produced but was not, for some reason. Obviously, this definition encompasses a vast array of possible reasons. Rather than attempt to record and act on every possible event or cause, loss accounting is a process of quantifying and capturing lost production by assigning it to loss categories that can later be aggregated and effectively analyzed to target problem-solving efforts at the biggest and best opportunities.
Choke points are steps in the value chain where production is often less than the capacity or full potential of the equipment. It's important to understand the difference between a choke point and a bottleneck. Most process engineers are familiar with the concept of a bottleneck, which is a physical constraint that occurs at the point in the process where the equipment with the lowest throughput capacity is installed. The bottleneck determines the overall maximum possible sustainable throughput rate of the whole plant—that's why process engineers look for ways to remove it. However, process units do not operate at full capacity all the time, so the bottleneck capacity does not explain actual production and is therefore of no use in reducing losses.
Unlike bottlenecks, there may be, and usually are, more than one choke point where losses occur, including but not limited to the bottleneck. To get started with loss accounting, a choke model is needed. A choke model is a diagram of the main choke points in the process and the quantified actual losses at each, usually based on a recent 12-month period. Figure 1 below shows examples of choke points. The first represents a mining process, which has a linear process flow. The second is a process called steam-assisted gravity drainage (SAGD), used in unconventional oil production, which includes a circular process. Note that each process has multiple potential choke points.
A choke model is an analysis of the actual losses that occur at each choke point. To do this, the actual throughput at each choke point is analyzed using historic data (usually hourly production data) over a period of 3-12 months, and reviewed with key people (operations, maintenance, planners, etc.) to determine the causes of lost production. Production and maintenance records may also be reviewed to shed more light on what was behind loss events. The reasons are then grouped into logical categories, including a bucket for "unknown" losses, which is interesting in itself. If the available data is insufficient to retro-actively determine the main causes of lost production then it may make sense to implement SIC on a temporary basis at the main choke points to determine what is actually causing the losses.
Every plant is unique, so the location and significance of choke points will vary. In one operation, the mine might be the biggest cause of lost final production. In another, it might be the grinding process. It depends on the capacity of the equipment, its reliability and how it is operated. The choke model is the basis on which a systematic performance improvement program can be designed. As illustrated in Figure 2 below, a good choke model shows the losses at each choke point and what caused them.
A choke model provides the facts needed to decide where SIC and loss capture should be implemented, and where the organization as a whole should then focus its efforts to gain maximum benefits. In the example above, it probably makes sense to install an SIC on the mine production and also one on grinding. However, there are other things to consider when deciding where to install SIC, such as available and accurate metering, boundaries of responsibility, and the location of stockpiles, which isolate one process step from another.
Choke models are closely related to the theory of constraints which is a methodology to identify the limiting factor in the process and then systematically improve that constraint until it is no longer the limiting factor. In lean manufacturing doctrine, the technique of line balancing is used to synchronize all steps in the production process to the rate at the current constraint. This makes it immediately apparent which step in the process is causing a line slowdown or stoppage at any point in time. However, this technique is difficult to achieve in large process plants, which have complex dynamics, and where temporary adjustments to production rates are avoided whenever possible. For these reasons, process plants tend to rely on intermediate product storage to absorb temporary imbalances. Any imbalance in inventories remaining at the end of a day is then corrected by the supply and operational planning process (S&OP) which determines the optimal production plan for the next day. Although not as conceptually simple as line balancing, this achieves the same goal of continuously matching overall production to the constraints.
Ideally, we would like to attribute losses in final production to losses that occurred at specific choke points. However, this is not as easy as it might sound. Typically, in a linear process such as the first one shown in Figure 1, chokes that are downstream of the most significant choke point will report frequent losses due to a lack of feedstock. Likewise, chokes upstream of a significant choke will report downtime due to downstream stoppages or lack of downstream storage capacity. In this case, it is not too difficult to see where the problem is.
In the case of a circular production process such as in the SAGD example in Figure 1, the processes at each choke point might be frequently shutting down due to lack of supply—shortages of steam for well production, shortages of produced water for water treatment, and shortages of boiler feed-water for steam production. In this case, it is unclear which choke (if any) is the problem. It might point instead to a serious production planning and scheduling problem, or a lack of intermediate storage capacity.
If losses are distributed evenly between two or three choke points, determining the causes of lost final production may not be trivial. If two process units coincidentally shut down during the same period, which one was the cause of the lost production? Rather than get mired in such debates, the goal of the choke model and loss accounting is to identify the key choke point (or points) and use the facts on losses to target initiatives to increase equipment utilization (OEE) at each.
Over time, the losses at each choke point will change. As losses are permanently eliminated at one choke point, they may increase at another. Therefore, the choke model will evolve and need updating over time. The focus of SIC and improvement efforts should also shift accordingly.
In the next article, I will describe the loss accounting process in more detail and how to use it to drive continuous improvement. Once a loss accounting system is in place, production losses are tracked as a matter of course and continuously guide management decision-making and improvement priorities.
What is the benefit of a choke model?
Identifies the points in the process where production losses occur
Quantifies the historic losses and identifies the main causes
Indicates where SIC should be installed
Organizational alignment on where to focus improvement efforts.