Maximizing alternative fuel benefits
with Expert Optimizer

ABB’s Expert Optimizer was first installed at Votorantim ­Cimentos’s Itau de Minas (Fig. 1) and Salto de Pirapora plants (Fig. 2) in 1995. Ageing hardware needed to be replaced (Fig. 3) and at the same time the opportunity to upgrade the optimization solution was taken. The challenges were to upgrade to the latest version without losing any of the existing plant specific features, and to bring the alternative fuels used in the plants into the strategy and therefore to bring them under better control. The upgrades were completed in 2009.

Founded in 1936, Votorantim Cimentos is one of the world’s top ten manufacturers of cement, concrete and aggregates. In Brazil it is the market leader, with 40 production units located in almost every Brazilian state, as well as 70 distribution centres and 90 central concrete plants throughout the country. It sells more than 40 products, including leading brands such as Votoran, Ita, Poty, Tocantins, Aratu, Votomassa, and Engemix. Outside Brazil it operates five factories and more than 120 concrete and aggregate units in North America. It also operates a factory in Bolivia and holds an equity stake in an industry ­leader in Chile, Argentina and Uruguay. The company has 11 000 employees globally and is considered one of the 150 Best Places to Work in Brazil, according to Guia Você S.A./Exame. It belongs to the Votorantim Group, one of the largest business conglomerates in Latin America, with industry-leading operations in manufacturing, financial services and new ventures.

Expert Optimizer version 2.0 was first installed at the plants in 1995 when the product was still known as LINKman. The upgrade to Expert Optimizer version 6.0 took place across two of Votorantim’s plants, Salto de Pirapora (Fig. 4) and Itau de Minas (Fig. 5). The situation at both plants for the installation of the latest version of Expert Optimizer is summarized in Table 1.

Votorantim was motivated to upgrade to the latest version of ABB’s Expert Optimizer at its Itau and Salto plants as the existing third party hardware was now 14 years old. The need to replace the hardware presented an opportunity to upgrade to the latest version of Expert Optimizer at the same time. The plant manager from the Itau plant Eng. Danilo Valadares Pinto was committed to the upgrade to the newer version of Expert Optimizer (Fig. 6): Ensuring cost savings, quality focus and reduction of manual interventions by the operator in the manufacturing process were vital objectives for the upgrade.

Two main challenges were presented at Votorantim’s Itau and Salto plants. Firstly, whilst upgrading to the latest version of Expert Optimizer presented an opportunity, it also raised some concerns for Votorantim. The sites had been gathering experience of and benefiting from Expert Optimizer since 1995, and during this time a great deal of knowledge about the ­system had been gained, and many site-specific features had been introduced. The auto-pilot approach had become a part of daily life at the plants. The operators knew how to ­interpret the information and how to use the existing version of ­Expert ­
Optimizer to help them achieve their goals and were com­fortable and confident with the current arrangements. It was essential for Votorantim that their operators’ knowledge and the site-specific features were not lost as a result of the upgrading.

Secondly, a range of alternative fuels are used across both plants. Variability in calorific value and tyres that often presented ­energy spikes needed to be accounted for in the new strategy, in order to maximize the benefits of Votorantim using alternative fuels.

Both plants were upgraded to the latest version of Expert ­Optimizer, version 6.0. The Salto plant’s upgrade was com­-
pleted in October 2008 and the Itau plant’s upgrade was completed in February 2009. Expert Optimizer is an Advanced Process Optimization (APO) solution. Advanced Process Optimization applications can be thought of as “auto-pilots” for the parts of the production process which they are configured to take care of (e.  g. kilns or mills) (Fig. 7). The best possible control strategy is implemented and continuously applied on a computer linked with the instrumentation and control system. The frequency with which the process can be analyzed combined with the application of advanced mathematical modelling, permits predictions about the process behaviour. Predicting the process behaviour is also combined with the knowledge which the optimization solution has “learned” from previous actions and outcomes. This enables the solution to make small, frequent, well-informed and tireless set-point changes which allow the process to be more stable than it is possible for a human oper-ator to achieve. Once the Advanced Process Optimization software has stabilized the process it is able to safely drive it closer to its operating limits and thereby maximize the desired goals of the plant where the optimization is in place.

The optimized limit to which the process is driven is defined on a case by case basis, according to the objectives of the business. Therefore it is possible to optimize for maximum throughput, for reduced fuel use, for maximum alternative fuel use, a combination of these, or many other options. It should be noted however, that Advanced Process Optimization is not a replacement for the plant’s normal instrumentation and control technology. Advanced Process Optimization is higher-level control system software that calculates improved operating points for the existing control loops.

Expert Optimizer version 6.0 employs a number of different methods to develop the Advanced Process Optimization strat-egies. The earlier version already installed at the plants was already using fuzzy logic, but the new version introduced the capability to use neural networks and model predictive control (MPC). MPC is based on the “receding horizon” principle. A sequence of optimum actions, which extend into the future, are calculated while incorporating the dynamics of the system in the time interval [t, t+p], in which t designates the current time and p represents a time horizon that is relevant for the application. In addition to the model, the cost functions are crucial in model predictive controllers for tuning the control behaviour. The optimization algorithm guarantees that the overall costs are as low as possible. The cost functions “translate” not only any deviations from the set point value but also any changes of an actuator into cost units. This allows a relationship to be formed between the variables. The variables have weighting and influence applied to their cost units according to defined objectives, for example, clinker chemistry, constraints on fuel consumption, market demand or the incursion of additional costs, if limits are approached or breached. This means that the control behaviour is achieved by intuitively simple cost functions and a sufficiently accurate model. This product is unique in the cement industry in having a list of several hundred completed applications in the fields of raw mix preparation, mill control, the production planning sector and last year completed its 200^th kiln installation.

As part of the upgrade there was a migration from the Alpha (Digital) workstations which were running with a VMS operating system, to a server running a Windows 2003 Server and workstations that use the Internet Explorer as the operator interface. All of the operator interfaces are web-based. This brought a friendlier and Windows feel to the interfaces, compared to the old version, while at the same time the structure and methodology of the application was kept.  Even with such a huge step, jumping from Expert Optimizer version 2.0 to Expert Optimizer version 6.0, the operators and engineers did not have a hard time stepping into the new version. They were able to use existing web skills, continue to use their existing site-specific features and system knowledge and only needed to learn about the new things that had been implemented into the system.

The site-specific information and configuration was changed which meant implementing a new strategy, not just copying the existing strategy, which provided an opportunity to install a simplified strategy.  This brought benefits in terms of the strategy being more robust and made things easier for the maintenance teams and the operators. According to the customer’s wishes ABB updated the PC kilns and all of the mills to the new ABB standard.  This included the installation of a new speed control feature which brings more stability for the circuits with a high efficiency separator.

Overall, knowledge that was implemented by ABB during primary and secondary service visits was retained.  Additionally, experience from the Votorantim side, which had been gained over several years, was kept whilst also being merged with the new features and functionality. This means that Votorantim’s operators are using their existing knowledge while simultaneously having access to the latest technologies incorporated in Expert Optimizer.

The energy in precalciners at both plants has been controlled with a rule-based approach since the first installation by ABB in 1995. The Proportional Integrated Differential (PID) control is not able to take an overall view of the master and alternative fuels and handles each fuel separately, therefore during the upgrade a new alternative fuels module was installed.The alternative fuels module considers the energy sources as a whole and so is able to handle set-points accordingly.

The module compensates with the master fuel (coal) the meas-ured deviation and also the non-measured ones, such as water content, calorific value and others. In the alternative fuels module consideration is also given to O₂/CO variations (Fig. 8) that are generated by the alternative fuels. In the precalciner strategy, gas coming from the cooler is also taken into consideration as a source of energy, in order to keep the precalciner temperature as stable as possible.

Due to frequent deviations, failures and or low reliability of alternative fuel resources, a “virtual fuel”, that modifies up to 5  % of precalciner energy according O₂ and CO measurements was implemented. Due to secondary air temperature oscillation it is also possible to manipulate the tyres current energy up to 4  % of the precalciner energy.

For example, at the Salto plant, the outlet temperature of the HOTDISC is accounted for in the strategy. This prevents the precalciner/cooler cycle (which normally uses alternative fuels) from becoming unstable, which is often triggered by an alternative fuels failure or variation, and if the corrective action is not fast enough or is incorrect, it results in disturbances that are hard for the operator to solve. The upgraded installation of Expert Optimizer can react every 10 seconds and so is able to control energy spikes which come in particular from the alternative fuel of tyres. The fast reaction time is therefore able to stop a negative cycle of feedback, before it becomes too difficult to recover without using extra resources and incurring high unnecessary costs.

Whenever the precalciner temperature function (Norm + Grad) is in the configured range the system keeps the average energy for the precalciner. The idea is to keep the energy as constant as possible and in a stable condition minimizing further disturbances due to setpoint changes. It also brings the energy setpoint back to the average in the transition conditions. A similar approach is used for the grate cooler control. Another benefit of this methodology (similar to the grate cooler control) is that it minimizes the frequency and the effect of potential cycles between the cooler and precalciner.

According to the previous trend Mr. Cassio Paiva Marcelo – Itau’s Expert Optimizer Specialist says “We can clearly see how stable the precaliner control is compared to the PID (Figs. 9 and 10), even when feeding mixed sorts of tyres. Thanks to its good performance and reliability the Expert Optimizer is online 95  % of the kiln run time”. Typical benefits that plants can expect to see when their alternative fuels are under Advanced Process Optimization APC control are in the order of:

­– Increased alternative fuel use (5  % – 20  %)

­– More stable product quality (10  % – 20  %)

­– Reduced fuel costs (5  % – 20  %)

­– Lower maintenance costs and fewer pollution emissions (5  % – 20  %)

Mr Francisco Andrade dos Santos, kiln operator at the Itau plant since before Expert Optimizer was ever installed, is already feeling the benefits of the upgrade. This is shown by his statement “After Expert Optimizer implementation, there was a big improvement, not just from the operational point of view but also in energy savings and production increase. If we have a problem with equipment we can take care of this and we do not worry because Expert Optimizer is doing its best all the time to keep the process stable”.

Votorantim has the possibility to keep more than 10 years of knowledge acquired from the previous version with these upgrades, plus the opportunity to have access to new ­technologies like Model Predictive Control (MPC) and Mixed Logical ­Dynamic (MLD) systems and other new applications developed by ABB. Since the upgrades were finished a host of benefits have been brought to the Itau and Salto plants. The optimization of set-points according to the quality analysis is helping to save energy by making small, frequent set-point changes. This means that drastic actions to bring the process under control, which would require costly extra energy inputs, are not needed. The stabilization of the process results in more consistent clinker quality and this in turn brings benefits to the mills. One of the main benefits that stands out is that both plants now have more stable use and control of the burning of alternative fuels, with less disruption to the process. This has allowed both plants to increase the amount of alternative fuels they use, which in turn is having a positive impact on reducing production costs. Expert Optimizer, part of ABB’s CPM Minerals suite of solutions, ensures that the kiln is operated under optimum conditions according to the customer’s goals. At the Itau and Salto plants this has meant that the energy spikes created by burning alternative fuels, especially tyres, are anticipated and therefore managed by Expert Optimizer.

www.abb.com/cement