Digitalization + AI Cut Scrap, Guide Smart Maintenance

Condals Group has turned its already sophisticated ferrous casting plants in Manresa, Spain, and Štúrovo, Slovakia, into next-generation digital foundries. Monitizer’s integrated suite - CIM, DISCOVER, DETECT, PRESCRIBE, and Trace and Guidance (TAG) products - now drive its processes with by automation, real-time monitoring, AI optimization, and traceability.

“We take decisions based on three main pillars: automatization, digitalization, and sustainability,” stated David de la Cruz, Condals Group chief digital officer. “Monitizer helps us to reduce scrap and increase efficiency, both of which improve sustainability. With these investments and our commitment to constant improvement, we have already achieved a 54.1% reduction in our overall scrap rate for AI-optimized, compliant castings.”

Automate and visualize production

Condals started by adding Monitizer | CIM modules to its three DISA molding lines to gather data, provide recipe management and process monitoring, and automatically control in-mold cooling time and casting sorting. CIM also synchronizes pouring with molding and mold transport to position the pouring unit’s nozzle exactly over each mold’s pouring cup for more consistent and accurate mold filling.

Next came Monitizer | DISCOVER, which collects data from any vendor’s equipment across multiple lines or facilities in any location, presents it for real-time monitoring in intuitive dashboards and stores it in a single, central database for analysis.

Supervisors and managers simply check their screen to see what’s happening in real time and make better, faster decisions. DISCOVER lets Carles Vives, production coordinator and purchasing manager, oversee every shift in every area of the foundry.

“I made a dashboard myself that shows supervisors the main parameters they need to monitor,” he said. “It shows the pattern that we are producing, if it has prescriptions, the cycle time, if the machine is working or not, the next pattern to produce, the level of the iron in the pouring machines, and so on.”

Lower energy consumption, smarter maintenance

To help cut energy usage, DISCOVER KPIs display live energy consumption for the melting furnaces and the Wheelabrator CT-3 shot blasting machine. Pouring furnaces will be connected next.

“There’s a fixed heat and time needed to melt iron, that’s our baseline,” explained chief quality officer Anna Castells. “Maybe we’ll see one furnace with lower efficiency, which prompts us to investigate and adjust it. Making the data visible gives us other opportunities, like adjusting shift schedules so we can avoid keeping iron hot while production is stopped.”

Monitizer | DETECT is also running at both facilities. Using the same interface and database as DISCOVER, DETECT constantly compares DISA equipment data against benchmarks. When readings fall outside tolerance - even subtle points that a human might miss - it triggers an alert with the likely cause and recommended actions.

Digital delivers intelligent remote support

In-house maintenance specialists use DETECT daily to monitor molding machines’ parameters during production, set alarm thresholds and make rapid, informed decisions. “You can check back to every previous pattern and run, and look at everything from hydraulic pressure and mold compressibility to the sand shot and the air exhaust,” reported Josep Lluís Victori, maintenance and safety manager.

Able to compare live data from similar machines across both facilities via DETECT and DISCOVER, the more experienced Manresa maintenance team supports the Slovakian facility remotely, helping to spot and solve potential problems. If the Manresa team needs help itself, DISA’s engineers are ready with online advice.

“There’s time pressure and stress if production has stopped, so it's very useful that another expert can look at your data and give you some hints on how to fix a problem,” Victori said.

Optimize production with AI

Monitizer | PRESCRIBE’s automated AI-driven analytics optimize an entire line, recommending optimal process parameters - “prescriptions” - that support stable, high-quality production to reduce scrap significantly.

Condals tested PRESCRIBE on one Manresa line. By 2021, it delivered an average 42% scrap reduction on two test patterns. Very impressive, but applying the AI’s intelligence in production and calculating ROI required changes to operational processes and company culture.

Condals found the biggest success factors for optimization were coverage: how many patterns have enough data for effective modeling; and compliance: how precisely operators follow the AI-generated control plan on their machines.

Driving AI success with coverage, compliance

“We need to produce a minimum number of molds for a pattern’s prescriptions to be statistically valid and reliable, which is harder to do with shorter runs,” de la Cruz explained. Extending PRESCRIBE to the second Manresa line increased coverage; today, there is sufficient data to model 80% of patterns.

Now, the focus is on compliance. Under 80% compliance for a pattern means PRESCRIBE’s suggestions are only partially implemented. This harms scrap reduction and breaks the self-learning feedback loop; if its original suggestions are “ignored” then the model cannot use the resulting data to improve its advice as it normally does.

“If the model is working as planned then, if we increase compliance, we should see a lower scrap rate,” he continued.

Empowering the workforce

To raise operator compliance, Condals installed large screens on each foundry floor where a single dashboard shows operators their actual setting value versus the AI-suggested control ranges. Regular meetings let everyone discuss recent results and give feedback on working practices that increase compliance.

“We had to incentivize and educate operators in how Monitizer supports their work and makes their lives easier,” de la Cruz said.

Discussions revealed that some prescriptions were too difficult for even the most skilled operator to meet. That’s because many machines just were not capable of sufficient precision to implement the narrow tolerances demanded. Now, the model “knows” that certain tolerances need to be wider for certain machines.

Monitoring compliance proves AI cuts scrap

“We have confirmed over the last year that, when we have high compliance, we have a low scrap rate,” de la Cruz reported. “We already run a good foundry with low scrap rates, so the improvements we see are small, but they’re consistent and always trending in the right direction.”

Over a period of six months at Manresa in 2025, average scrap rates were 51.8% lower for Line 1 and 56.4% lower for Line 2 when compliant with prescriptions. Condals also learned that, even with lower compliance, encouraging operators to stick to setpoints reduces process variability and improves results.

PRESCRIBE implementation is about to commence at Štúrovo, and Condals recently deployed a casting-level tracking system. TAG collects individual process data for each casting to increase accuracy and further improve AI modelling.

Partnering with foundry engineering experts

Condals now monitors more than 152,500 data points in its relentless drive to improve production. “We have learned a lot about how important governance is when you apply an AI model and I think that is rare in our industry,” David de la Cruz said. “From data quality to transferring learnings to the operators and following everything up in our monthly reports, how you manage all these things is critical.”

“Condals recognized data as a key asset and we are changing company culture to embrace this new era of digitalization,” CEO Eduard Salles concluded. “We needed a strategic digital partner to support our digital transformation, and Monitizer’s technology and foundry expertise have delivered significant added value for us.”