CSense is a Necessity in an Economic Downturn - Wednesday, August 19, 2009

There are two rarely acknowledged benefits of an economic downturn; it's a wakeup call to cut production costs and it gives you more time to think about how you're going to do that cutting. Just a year ago, cutting production costs was the first item on the agenda – now it's the only item on the agenda. And the old adage of being too busy to try something new can no longer be used as an excuse because a slowdown, no matter how modest, gives you the ideal opportunity to implement improvements so that you'll be ready for the inevitable upturn.

According to Warren Buffett, possibly one of the greatest investors of all time, "Be fearful when others are greedy, and be greedy when others are fearful." Since the Great Depression of 1929, there's never been a more opportune time to be greedy and to want a bigger share of the pie when the economy turns. It's Buffett's opinion that "...most major companies will be setting new profit records 5, 10 and 20 years from now."

But they're not going to get there by standing still. Successful businesses will adapt to the new economic situation and they will adjust their strategies to capitalise on these new realities. Some of these strategies will include renewed and aggressive cost-cutting initiatives that will preserve valuable plant and human assets while maintaining quality, increasing throughput and improving their agility to cope with fluctuating market demands.

Hazardous processes, excessive maintenance, inefficiencies, process malfunctions and poorly understood or badly controlled processes are all major contributors to increasing production costs. Fortunately, it's possible to deal with such scenarios through the early detection and elimination of their root causes by using troubleshooting, smart process monitoring and sophisticated control systems that include advanced process control (APC) technology.

The CSense Process Performance Enhancement suite is being successfully deployed in many industries and applications to address the issue of reducing production costs, improving safety and empowering operators and production management. So much so, that a level of effectiveness was realised in each case which would was not possible with conventional statistical control methods. Here are a few examples:

Accident-prone processes
Reduced sensor and maintenance cost
Efficiency? By all means – but at what cost?
 Production glitches
Uncontrollable variables
Cost complexity

1. Accident-prone Process

Some processes inherently have the potential to lead to dangerous situations, which would have severe safety and cost implications if the worst was to happen. Such is the case with Teckcominco's Trail operation, which is one of the largest integrated non-ferrous facilities in the world. They use a progressive process to recycle zinc residue through a lead smelter. After flashmelting the residue, the lead bullion is tapped from the bottom where it settled. However, the lead is tapped through a tap block which is water cooled. The severe consequences resulting from water coming into contact with superheated lead are obvious and must be avoided at all costs. The company called in CSense who installed a system which would prevent such situations. Principal component analysis models, heat load models and the generation of "Crisp rules" determine the health of the tap block and warn the tapping operator when the situation becomes dangerous.

"Based on the work done to date, we have a far better understanding of the tap block and how it reacts," says Robert Zwick, superintendent of Process Control. "It's certainly a complex problem but now the system is providing value by giving an indication to engineers and operators as to the integrity of all our tapping blocks and when they may potentially become unsafe to use. Further enhancements are scheduled to provide us with greater confidence in the system and a more reliable health index indication."

2. Reduced Sensor and Maintenance Costs

Excessive maintenance on a plant equals high running costs. An example is the frequent servicing of sensors, which are often placed in hazardous environments and subject to failure. In a worst-case scenario, an entire process is halted when a sensor fails, as is the case at the De Beers Venetia mine. This mine's ore beneficiation process is subject to feed with varying characteristics, making weightometers and densitometers critical to the understanding of the state of the process. When they fail, the plant is stopped, costing De Beers as much as half a million Rand an hour. CSense designed soft sensors for this plant, replacing or monitoring the real sensors. Models (fuzzy logic, expert systems and neural networks) of the process predict values for the real sensor, allowing the process to continue in the case of a failed sensor. The soft sensor tons/hour readings have a greater than 90% correlation with the actual sensor readings. What's more, since weightometers and densitometers are expensive, soft sensors can be used where there are no physical sensors at all. "CSense and soft sensor technology help us measure accurately even where no sensors are available. Strange? Only at first," says Xiaowei Pan, Senior Process Engineer, DebTech,

3. Efficiency: By All Means - But At What Cost?

As all project managers on a plant know, a more efficient process could mean increased operational costs, negating the advantage of increased efficiency. This is especially the case when the effect of individual strategies is not known and some may not contribute to efficiency, but only to costs. Umicore decided to eliminate unnecessary costs at their Hoboken electrowinning plant and managed to increase efficiency from 80% to 90%, but at a significant increase in operational costs. The effects of some of these new measures were not known with the result that some had the reverse effect than that intended. With the help of CSense models simulating cause and effect scenarios, efficiency was raised to 94% (without any corresponding increases in cost) and tens of thousands of Euros were saved.

4. Production Glitches

At Saldanha Steel an important production problem is cobbing, when steel coils 'jam' between rolls. This leads to wasted productivity and revenue, because all affected steel coils and upstream slabs are scrapped. Before CSense was called in to sort out the problem, 0.4% of all steel coils were cobbed. Within 30 minutes the cause was identified as a pressure transducer on the side guide. When this transducer was replaced, the root cause was eliminated as was the cobbing.

5. Uncontrollable Variables

Sometimes a certain variable has a plant-wide impact and its control becomes critical. At Impala Platinum, poor pH control in their base metals refinery led to higher concentrations of base metals such as nickel and iron, which negatively influenced efficiency in the platinum metals refinery (PMR). In extreme cases, compounds contaminated the platinum group metals solids, necessitating recycling of the entire batch. Thus, the pH control upstream has a great impact on the profitability of the PMR process. In this process, numerous independent variables affect the pH making it difficult, if not impossible, for operators to run the process optimally. However, with an APC solution from CSense, all of the complex process dynamics are taken into account. The result was a 40% improvement in pH stability linking to an increase of 1.4% in the PGM grade. "Although these numbers may seem small," says Tim Spandiel, Manager, Impala Base Metals Refinery., "they are extremely significant to the financial benefit of all our refining processes."

6. Costly Complexity

Part of the smelting operation at Lonmin's facility in Marikana involves drying platinum group metals concentrate through filtering and flash-drying processes – which were identified as production bottlenecks. The solution proved to be more difficult than anyone had thought and required the help of an expert system that could help operators effectively manage a complex, multi-variable environment.

Checking on the variances between the filter plant with the drying section led Lonmin to the conclusion that they were working against each other and that the only way to improve efficiency was to integrate them into a single entity. The complexity of running such a unified process optimally while juggling multiple interdependent variables is outside the scope of manual control.

"We managed to integrate two conflicting processes without spending capital or changing the plant in any way. This CSense-based system also helps operators to understand the process and to change it as necessary," says Frans de Beer, Senior Manager, Smelter.

7. Conclusion

For mining companies and manufacturers, production processes are where their wealth is created. To that degree, they should be optimised by whatever means possible because an improvement at this core level has a huge leveraging effect on the profitability of sold products. In many instances, such as the metals industry, the final product price is pegged by market demand, meaning that any increase in profitability can only be achieved by reducing production costs.

Quite often, production efficiency improvements and cost reductions are achieved without adding capital plant or costly control systems but by using the information already available in SCADA historians and mapping this to accurate plant models and diagnostic software. In all cases where CSense's solutions have been implemented, the unanimous comment from end-users has been that they now understand their complex processes a lot better – knowledge, after all, is power.

So, process performance enhancement is no longer a luxury but a necessity for companies who want to weather the current financial storms and capitalise on the opportunities that will inevitably follow - and the right time to implement these initiatives is when there is breathing room to think clearly and "out of the box".