Manufacturology

Process cost modelling in steel industries: Flexibility, flow and production cost

I’ve writ­ten about how process flex­i­bil­ity and small batch sizes improve pro­duc­tion flow in steel indus­tries in a cou­ple of ear­lier posts (here and here). In this post I’ll dis­cuss how the eco­nom­i­cal con­se­quences of flex­i­ble rolling mill tech­nol­ogy can be stud­ied using a con­cep­tual process cost model.

The fig­ure below illus­trates a con­cep­tual cost model that I’ve been using to illus­trate the effects of improved process flex­i­bil­ity in hot rolling. I’ve pre­sented vari­ants of this model at sev­eral con­fer­ences. Ref­er­ences are found under pub­li­ca­tions.

The level of process flex­i­bil­ity bal­ances costs for work rolls, WIP and reheat­ing energy in a hot strip mill.

This con­cep­tual model illus­trates how the level of process flex­i­bil­ity bal­ances the costs for work roll con­sump­tion against buffer­ing and reheat­ing energy. A num­ber of para­me­ters are influ­enced. These are rep­re­sented by text labels encir­cling the outer ellipse.

As I’ve writ­ten before, process flex­i­bil­ity yields the capa­bil­ity to process an arbi­trary sequence of prod­ucts with min­i­mum time and cost penalty. It serves as a “buffer against vari­abil­ity” by increas­ing the short-term abil­ity to process an unplanned sequence or com­bi­na­tion of products.

In steel plants, process flex­i­bil­ity deter­mines (at any given time) the capability

• of the melt­shop to pro­duce a par­tic­u­lar steel grade;
• of the con­tin­u­ous caster to cast a par­tic­u­lar steel grade and slab geom­e­try; and
• of the hot strip mill to roll a slab of a par­tic­u­lar grade, width and thick­ness into the desired tar­get thickness.

Unfor­tu­nately, process flex­i­bil­ity improve­ments can appear to yield reduced pro­duc­tiv­ity and increased costs as seen from the per­spec­tive of an indi­vid­ual process step. The over­all pos­i­tive effect is on the sys­tem level, and thus less obvi­ous as the cause and effect are sep­a­rated in both time and space.

When I made this model, I had found that the most impor­tant cost dri­vers were WIP, reheat­ing energy and roll wear in the hot strip mill:

(a)  Reheat­ing energy: If lead­time and buffer­ing is reduced, some heat from the melt­ing and cast­ing is pre­served, and the mean tem­per­a­ture of slabs enter­ing the rolling mill increase. Since slabs must hold about 1250°C dur­ing hot rolling, and reduced lead­time allow fuel con­sump­tion in the reheat fur­naces to be lowered.

(b) WIP: The amount of WIP and hence the cost of cap­i­tal tied up in in-process inven­tory is reduced. This is of par­tic­u­lar inter­est to pro­duc­ers of stain­less steel due to high raw mate­r­ial prices.

©  Work rolls: Increas­ing the num­ber of roll changes also require more fre­quent con­di­tion­ing of the work roll sur­face. This may result in raised over­all roll con­sump­tion, hence caus­ing the tool costs to increase.

I used this model as a basis for a sys­tem dynam­ics model and used some opti­mi­sa­tion tech­niques to deter­mine the char­ac­ter­is­tics of the model. I was then able to get the results indi­cated with “basic model” in the plot below.

Cost curves pro­duced with a sim­ple con­cep­tual cost model and a more advanced dynamic process cost model.

These results show how reduced setup time in the rolling mill will ini­tially lead to reduced over­all costs. How­ever at some point, the cost curve turns upward because every time the rolls are changed, they are con­di­tioned in a roll grind­ing machine. If the rolls are changed very often, it becomes impor­tant that the roll con­di­tion­ing is done with a cer­tain pre­ci­sion that pre­vents unnec­es­sary removal of fresh roll mate­r­ial under the dam­aged sur­face layer. The “basic model” could not account for this since con­di­tion­ing was always assumed to be “inefficient”.

The next step was that I devel­oped a more advanced model which I used to look at what hap­pened when more adap­tive roll grind­ing tech­nol­ogy was used in com­bi­na­tion with rapid roll changes. These curves are indi­cated in the plot as “dynamic model”. Basi­cally these curves illus­trate the obvi­ous fact that “pre­ci­sion” roll grind­ing that remove lit­tle excess mate­r­ial from the rolls yield bet­ter over­all econ­omy than if you waste the roll mate­r­ial by using an inef­fi­cient grind­ing process.

The next step is to account for prod­uct vari­ety. This com­pli­cates things to the extent that I used much of my PhD the­sis to model and analyse this prob­lem. How­ever, the sim­ple model shown above is still very use­ful because it shows how changes in the degree of process flex­i­bil­ity influ­ence the bal­ance between cost drivers.

I’ve argued before that process cost mod­el­ling can help indus­tries to become more inno­v­a­tive since they can be used to show how process flex­i­bil­ity improve­ments yield high level strate­gic capa­bil­i­ties that com­peti­tors do not pos­sess. When a process cost model is devel­oped using sys­tem dynam­ics or some sim­i­lar sim­u­la­tion approach, it becomes pos­si­ble to see dynamic effects that are oth­er­wise very hard to quantify.

Be Socia­ble, Share!

•