The integrated steel mill of Isdemir is located in south east Turkey, nearby the city of Iskenderun, and is part of Erdemir Group which is the leading supplier of flat steel in Turkey. Besides Isdemir plant Erdemir runs a second plant at Eregli.
Originally a long steel products manufacturer, Isdemir plant started producing flat steel in 2008, with the acquisition of two continuous slab casters and a hot strip mill with a 3.5 million tons/year hot rolling capacity.
In 2010, one year before Go-Live of the new Advanced Planning environment, the hot charge ratio was 5 %. To increase this figure and to materialize countable benefits Isdemir focused on decreased energy consumption in the slab re-heating furnaces, the ability to generate hot strip mill schedules with a lower slab stock level, and a decreased number of slab movements, with a positive impact on crane activity. These measures should finally result in shorter lead times. Erdemir Group and PSI worked out an integrated planning concept that should allow reaching these goals. Today, 4 years after Go-Live of the system, it is worth to have a close look on this project and its success story.
As part of a continuous effort to reduce costs, increasing the ratio for direct and indirect slabs hot charging has become a top priority for most integrated steel mills. Those processes consist in charging casted slabs in the rolling mill’s reheating furnaces while they are still hot, which allows for substantial energy savings as well as reduced inventory and slab handling costs. The integrated steel mill of Erdemir Group at Isdemir started working with PSImetals Planning in 2011 and has achieved great improvements.
The key objective in this overall production planning optimization was to maximize the slab hot charging ratio between the slab casters and the hot strip mill.
Whenever possible, steel producers indeed try minimizing the energy needed to reheat the slabs at their rolling temperature (1200°C) by charging those as hot as possible in the reheating furnaces in front of the mill. This is feasible if the slabs are rolled directly after being casted (direct charging) or within a period of up to 12 hours (hot charging). The gas savings that can be achieved in the furnace allow for substantial cost savings.
However hot/direct charging is not always possible as certain steel qualities require a mandatory cooling of the slabs between the casting and the rolling processes. For certain applications, a mandatory scarfing of the slabs is also necessary. On the other hand certain steel qualities may require mandatory hot charging: these steel qualities need to be charged as soon as possible after the slabs are produced.
In the mid-term horizon, the objective is to try identifying future hot charging opportunities based on the order book demand, and to prepare synchronized HSM and casters rough schedules accordingly. The detailed scheduling on the short term horizon will then try to concretize these opportunities based on the actual production lines and material situation.
Both midterm and shortterm layers are updated on a daily basis based on the latest production situation. Therefore the planning department of Isdemir loads every day the production order book for the upcoming months into the planning system, as well as the work in progress (WIP) stock units in front of each line (i. e. all existing slabs, billets, coils) and the planned line downtimes. Those dynamic inputs are uploaded from the production execution system in place at Isdemir, named UYS, on a snapshot basis (every night, or on demand), and the initial stock units have been previously assigned to production orders.
Increasing the hot charge ratio is not possible by only focusing on the isolated schedules of caster and hot strip mill. Therefore PSI hot charging solution combines the midterm horizon (up to 4 weeks) with short term schedules (up to 3 days).
Increasing the hot charge ratio is not possible by only focusing on the isolated schedules of caster and hot strip mill.
On the midterm horizon a daily flow plan is built to find the best tradeoff between service performance, throughput maximization and inventory levels minimization, while also calculating optimized rolling diameter campaigns at the wire rod mill. These results are provided as targets to an order scheduling process, which generates rough cross-line schedules at order and piece granularity over the upcoming 4 weeks. This process allows simulating the expected production dates for each order on each line, expected completion dates, and expected inventory levels. High level technical constraints that can generate due date dispersion for periods larger than a few days, such as backup roll campaigns at the hot strip mill, or section campaigns at the billet casters are also taken into account at this stage. Furthermore, the schedules of certain lines need to be highly synchronized. In particular, billet casters need to properly feed the wire rod mill campaigns and the slab casters schedules need to be synchronized to the hot strip mill schedule to maximize hot charging opportunities.
Finally, the detailed scheduling process aims at generating optimal detailed schedules at piece level for each production line for the next shifts up to 3 days, considering all the lines technical constraints such as grade transitions, width jumps, etc., and trying to respect the target order production dates received from order scheduling. The output of this process consists in schedules ready to be sent to production.
The key to success is the identification of hot-charging opportunities on the midterm horizon through a pullpush synchronization of casting and rolling schedules.
In the midterm horizon Isdemir starts by generating an ideal HSM sequence, the ‘pull’ run, which will:
As a result of the HSM pull run, a maximized number of hot charge coffins have been scheduled.
It should be noted that at this stage the number of WIP slabs is not sufficient to build a midterm HSM program covering 3 to 4 weeks, and since caster sequences haven’t been created yet not many forecasted slabs are available to complete the HSM sequence (only those corresponding to the frozen ‘running production’ sequences at the casters). To address this, the slab yard population is completed by generating virtual slabs.
The next step is to pull these ideal HSM requirements to the casters, and try generating caster sequences that fulfil them at best, while considering casting scheduling restrictions.
This means that the local due dates for caster scheduling will correspond to the hot charging needs of the HSM, thus maximizing the chances to realize the identified hot charging opportunities of the pull run. It is important to note that the original order due dates have of course been considered as part of the HSM pull run constraints, but there exist a trade-off between satisfying the due dates of individual orders and the order grouping into hot charge coffins.
Since the casting limitations have now also to be taken into account, it is probable that not all hot charging coffins can be fed exactly as required by the pull run. This is why the HSM schedule will be finally re-calculated in a ‘push’ run, this time considering only real slabs and the forecasted slabs from the casters. At this stage, the previously created virtual slabs have been deleted. By definition, the virtual slabs were directly available for scheduling; however the forecasted slabs have a precise arrival time based on the casting schedules. A particularity of the HSM push run is that it tries keeping the coffins selected during the pull run, but will now re-optimize the sequencing and filling of those coffins with the forecasted slabs instead of the virtual ones.
In the short term horizon, the planner must prepare casters and mill schedules for the upcoming shifts, up to 3 days. Once planned, the short term schedules will be released to UYS. Hence, the sequences must be more detailed than the rough schedules generated at the order scheduling level, considering the full set of technical constraints, such as specific jump transitions between slabs. Some detailed constraints are indeed not considered within the midterm optimization for performance reasons, as well as due to the fact that the exact production order and material picture is uncertain beyond the next few days. However, the goal is of course to materialize in the short term the hot charge coffin opportunities that have been identified within the higher planning level.
The hot charging optimization in the short term is done in a pure push mode: at first, optimized caster sequences are generated, considering the imposed mould campaigns and tundishes for hot charging.
In a second step, the planner generates a short term HSM sequence. The coffin selection is not done automatically at this stage, but manually by the planner. By default, the coffin selected by the midterm push run is proposed; however, the planner may select another coffin depending on the actual situation at the caster. Indeed, especially for hot charge coffins, it is critical to verify if the short term caster sequence can provide in time the necessary forecasted slabs to fill the coffin.
The propagation of forecasted slabs from the casters sequences to the stock in front of the HSM is at this stage directly handled by Isdemir’s UYS. As soon as the slabs are effectively casted and come into existence, UYS is also responsible to transform those forecasted slabs into real slabs.
Five years after the implementation of hot charging optimization at Isdemir based on PSImetals solution, the customer could gain the following benefits:
The reduced energy consumption is directly linked to the gas savings at the reheating furnace, that could be achieved thanks to a higher hot charging ratio (hot slabs require substantially less reheating to reach their rolling temperature). The shorter lead times and reduced slab movements, as well as the possibility to work with a globally lower slab inventory are explained by the fact that hot slabs recently casted are just stored in a specific area of the slab yard for a short period of time, and less slabs need to be taken from the cold part of the yard. Finally, charging already hot slabs in the reheating furnace allows throughput gains by working at a faster furnace pace.
Finally, charging already hot slabs in the reheating furnace allows throughput gains by working at a faster furnace pace.