Scheduling Job Shops and Engineer-to-Order – Three Alternatives that can Work

Job shop production is a challenge for scheduling.

Here are three ideas for job shop scheduling and scheduling for Engineer-To-Order and customized products.

Each of these ideas can be realized with an automated Excel tool.   Give feedback in the comments, we can develop a tool for you and add it to the download page.

Job shops and other suppliers of customized products need production scheduling.  Demand can be more uneven and unpredictable than repetitive manufacturing.

As well as the uncertainty of which product will be in demand, there is also uncertainty of what materials it will require and how much work is needed from which work centers.

Products that are highly customized usually carry critical due-dates.  Customers depend on getting the order on time – they often do not have an alternative supply.

A simple truth about capacity planning is that predictable production performance arises from a simple calculation.  This compares minutes and hours of work in a product with hours of available capacity.  A process routing is the starting point, yet Engineer-To-Order products have no standard process data.

How to generate routing data inside the fulfillment cycle?

There are three alternative approaches to job shop scheduling, scheduling customized products and any other manufacturer who have no standard process data.

1.Scheduling Engineering in Pre-Production

In highly engineered systems, there is usually a lot of design and product/system engineering before work is released to manufacturing.

The order confirmation point can sit anywhere on a line that stretches from defining the solution to production.

The main application for scheduling is production.  But why should we limit it to this?  In many cases, engineering work that goes into a solution is value-added.  It requires skilled, finite resources and is under time pressure.

Integration between engineering and production leads to predictable delivery and cost  performance.  There is no  justification for silos between sales, engineering and manufacturing.  In this view, engineering can be seen as another work operation that feeds production.

The scheduling of pre-production generates routing data for production.  Engineering outputs are dependencies.   The product of upstream engineering  is an input for work centers at downstream production.

In some cases, this is not just analogy.  Bills of Material can include documents and certificates that are required to ship or release work into manufacturing.  QA/QC and regulatory documentation generates before and during the manufacturing process.

Before engineering builds scheduling data for production, you need to start with some token data.  Process families give estimated times that support an initial scheduling run.  Then when engineering operations are completed, it revises the scheduling for downstream production.  Estimate assumptions should be checked carefully, else this feedback loop gets unstable.

Most job shops and ETO manufacturers have order configuration software.  This system is usually connected to the generation of drawings and Bills of Material.  Less common is an integration to routings and process definition.  Fewer companies truly integrate work-flow between engineering and production.

If you did do this, imagine the benefit?  Faster Request-for-Quote (RFQ),  better capable-to-promise, management of engineering constraints, eliminated silo-thinking and even (dare we say it…) design-for-manufacturing and concurrent engineering!

2.  Job Costing and Configuration

A good costing process has the same granularity as capacity constraints.

What do I mean by this?  If you have constraint work centers, then job costing should drill down to hours and minutes at each of those work centers.  Any constraint to revenue and cash-flow needs to be properly sized up in the cost of sales.

So it should be simple, right?  The job costing process already generates the level of precision that you need for process routings.

In practice, configuration and costing do not always align with process definition.  They should do but they don’t.

The obstacle may be technical (system doesn’t support it), accounting (cost pools too broad) or organizational (silos, overly optimistic costing assumptions).  Either way, it could be that the order configuration process needs to be thrown out enhanced and highly specific calculations on cost and capacity made to fit.  By the way, spreadsheets are great for doing this.

It doesn’t have to be an all-or-nothing process.  Improve configurator inputs, enhance the logic or extend the functionality with an open, Excel-based system.

And measure estimates against actuals to constantly improve the process.  A closed feedback loop of measurement can deliver twin improvements in higher costing accuracy and better production capability.

3.  Automated Calculation from Master Data and History

This option is trickier, but offers strong advantages based on a scientific approach.  The premise is that process routing data be derived from order parameters.

This means that given a BOM and some values from configuration, you can calculate the routing result.

Now clearly this depends on the complexity of product engineering.  Manufacturers of products customized from a base model have a good chance of making this work.  Suppliers of highly engineered solutions will not want to fully automate this process.

We can simplify the process by breaking it down into constituent parts.  A process routing for a job shop (or anyone else) comprises of work sequence and content.  The sequence of operations that each product requires is a starting point for a process family.

In many job shops, there are a limited number of work center sequences.  Putting aside the actual time at the process and comparing only the operation sequence, new products will fall into one of a manageable number of process families.

Manageable means dozens at the most, not hundreds or thousands.  Interrogate the order history and see how many different sequences have been produced in the last few years.  What relationships there might be between product family and unique sequence?

Is it possible to select the right process family from a list at order entry?  Maybe it could be done by process of elimination.  This is a nice application of the suggestion of smart drop-down boxes that can be powered by the INDEX function.

Once the operation sequence is set, the next step is work content.  This is trickier and requires a flexible configuration tool that can handle many variables in a linear calculation.

In theory, all work times relate to order parameters like this: ax + b = c.   is the volume, a and b are parameter values.  To state this another way, all standard times have a variable component and a fixed one.

Job shop and engineer-to-order manufacturers regularly accept demand for products that they have never made before.  This rarely means that they accept demand for product with no similarity whatsoever to previous orders .  The production and order history can show correlation between order attributes and production requirements .

Data analysis can discover adaptable rules that link order parameters with a prediction of resource requirement.  When put together with the material structure in the BOM, it is possible that information exists to make calculated predictions of working time.

Try it, test it and see if any patterns emerge that are useful to the scheduling process.

How to Put These Ideas Into Action

The theory is nice, but how do we make it happen?

Our job shop and engineer-to-order clients often come to us with configuration, costing and production capability problems.  We solve these problems with real Excel-based solutions based on their business requirement.

You understand that we can’t show you client examples.  but we can do the next best thing.  If you let me know which of these ideas fits the best with your job shop scheduling or Engineer-To-Order business, we can build a demonstration system for you.

This will be free from confidential client data, but a true example of what works in the real world.  We might even give away a download version to our subscribers.

This is what you have to do.  Firstly, if you have not already done so, sign up as a subscriber for free.  This will open in a new window, so you can do it now.

Next, tell me in the comments which of these ideas for building routing data fits the best with your business:

  1. Schedule Engineering in Pre-Production
  2. Job Costing at Configuration
  3. Automated Calculation from Master Data

Give us more detailed comments and help us give you something of value.

38 thoughts on “Scheduling Job Shops and Engineer-to-Order – Three Alternatives that can Work”

  1. hi, great blog and info.

    for me it needs to be job costing. we have a configurator but cost pools are too general. how to tie a spread sheet into our system, I dont know. lke to see what you have got in this area

  2. It would be nice to calculate this from order details. But in practice we need intervention from manual engineering to prepare the work order, so scheduling engineering tasks would fit the best for me.

  3. Great web site! I’d very much like to see something on the automated calculation from order data as it’s the primary direction we’re heading based on a machine capability matrix.

  4. I would like to see some suggested solutions for our company that deals with ETO solutions for food processing companies. I am involved in the costing and estimating side of sales so and example of Job Costing at Configuration would also be beneficial.

  5. Enrique Romero

    Pretty interesting, but how can we handle those job shop and ETo situations in presence of setups??, something like minimizing setup cost???

    Thank you

    1. Hello Enrique,

      in the case of minimizing setups, the total setup time is sensitive to sequence, correct? Setting up to go from A -> B takes longer than A->C because B and C are different part families. A paint line is a classic example (grey-to-black quicker than grey-to-white).

      the usual way we handle this in repetitive manufacturing is with a setup matrix (or table) that defines the setup time from each part to every other part (or family to every other family).

      We can then run a simple rules to define the sequence in such a way as to limit the average setup time. The easiest is to group work orders for like parts that come up in the same time interval, usually a day or two. This changes the sequence set by due-date, but you know the interval and you can control due-date performance.

      Setups in ETO can be handled the same way.
      The difference is that instead of defining each setup from part to part, you do family to family. Or you break it up into components – what work is fixed (same for every setup) and what work is variable (against volume for that part family).

      This is all good data that really helps a SMED effort. In fact, the measurement and control that scheduling gives you in modeling setups in is one of the most powerful tools for SMED and setup Kaizen. It provides a very effective tool to project benefits in terms that management will understand.

  6. The job costing and configuration is probably must useful for my job shop environment. I have found that labour is the contraint rather than work centres as the tradesmen are not tied to one machine or area and can often share work areas. To then tie in job costing to my current excel schedule would be great

  7. As we are manufacturing all customized control panels and evertime we have put all the data viz. bom, mi, purchase followups, etc this would help us out to be better…

    1. Not necessarily. If your costing process is creating the right data, then it would make sense to tie planning and scheduling to this information.

      If the costing process does not provide you with the right level of granularity then you can always generate it with engineering instead or derive from other data not directly used in costing.

  8. A good costing process has the same granularity as capacity constraints.

    This is the most important sentence you have written. The big problem of course is the stability of this constraint. A software models a particular situation. Due to the varying input of products, the bottleneck changes as well. So how well will the software describe the actual situation?

    1. Thank you for your input. I agree that costing is a critical dimension to this discussion.

      If the costing process has the same granularity as constraints, this doesn’t mean costing is limited to constraints. It can include all operations at the same level in the hierarchy. Do them all and then you have the data you need for both costing and order execution.

      After all, a commodity part like a washer may not be considered a material constraint but it still appears in the BOM. It adds up in the total material cost. If you miss it out, you can’t ship the product. Why not the same for non-constraint operations?

      Good manufacturing practice has symmetry between material and process definition. A BOM and a Routing are two elements of value definition. Many production execution problems come from a bias towards materials and product at the expense of process visibility.

  9. I am not sure,, for me is better idea to have standar time for each family of products and performe rccp with this info. When i have worked with Historical, or also with the time calculated by Engeneering it always has a bid deviation that reflects in bad shop floor performance

  10. Schedule engineering in pre-production would probably work best for us. We have no routings and purchase most of our processes outside. Production is always delayed due to engineering and the lack of pre planning. My goal is capacity planning, production scheduling and reducing cycle times.

  11. I work frequently with job shops to help then recover from past due situations. I have multiple orders with varying due dates, that compete for machining resources. I would like to be able to input existing router data, order quantities, and due dates, then calculate work center loads, and best order sequencing. Once this is complete, I would like to provide a report out as a gantt chart, showing only those orders I am tracking, and the dates each order will finish each operation in it’s router. All operations could be listed in a row with all jobs and part numbers listed in columns a dn b. The intersection of these would contain a date.

  12. Declan McTiernan

    I like the ideas put forward here. I am in charge of projects costing at our company so I am now working on developing a costings/configuration solutions as suggested.

  13. Luis Carlos Jaramillo

    In my company we get WorkOrders to build or repair items that are completly different (no costumers ask for the same service, and it´s often a 1 time deal or some a year), and we have a fixed amount of workstations. (5 diferent processes, 17 machines). The WorkOrders are stacking up, and we are having trouble keeping up with our duedates. Would it be Schedule Engineering in Pre-Production, if i need to organize the secuence of each WorkOrder operation at each workstation, to meet all of my DueDates?

  14. I am an advisor in a pharma lab. I am interested in scheduling our Quality Control laboratory. The cycle of liberation of products can take days. Reducing this time is one objetive.

  15. We are a custom shop with numerous scheduling issues because of the high demand and continuous demand for our product. “A problem everyone wants to have right”. I would love to have a consistent way of scheduling work so that parameters such as personnel competence, machine capability, capacity, among other variables can be addressed. Tired of missing due dates. We are very effective but our efficiency needs help.

  16. Scheduling in a HMLV environment is hard, I work as a planner for a company that assembles to order and engineers to order, for projects (engineer to order) every new part number is loaded into the system with their respective BOM, we have about 164K different part numbers which later on are assembled to order once the customer starts ordering spares for their projects now at site start needing maintenance, how to integrate both of the planning methods is still a question and something we have quite an area of opportunity for us planners.

    The main problem we have is we don’t have standard working hours we can work on, but production supervisors usually have the answer to how many days will it take, and for buyers, their vendors usually give them pretty accurate lead times, once the project’s SO is loaded, we can start working on creating the WO’s and releasing them to the warehouse for picking materials and sending them to their specific Job Shops (Machine Shop, Weld, Mech. Assembly, Electrical Assembly, etc.)

    1. My question here is how to make a production schedule efficient enough to give accurate promise dates, we have 95% OTD, but this is not good enough since the promised dates differ a lot from required dates and we end up finishing labor before the promised days (1 to 5 days in average, and sometimes even weeks ahead) and to the 5% miss on promised dates usually refer to vendor delays, and there’s no way to expedite labor since the production supervisors start working on whatever is available without considering what is going to arrive from suppliers throughout the day, even warehouse picks material upon arrival, and we have wasted a few setup’s having to expedite work orders to finish on time that one order that got delayed instead of finishing early what has already it’s components on hand.

  17. Hi. Any ideas how to implement an excel based finite scheduler that is linked to an existing ERP (JDE Enterprise One)? Products are ETO and high volume, high mix.


  18. Like Sales-Prod, we do a lot of repeat work that may be variations of a base part. We also do ETO with extremely short lead times. Sometimes only a day or two. I am looking for any way to schedule. We currently just use an EXCEL spread sheet as a “priority list” with just basic information, Part #, description, quantity, due date.

  19. by the way you explain it all, it seems Automation Calculation is it. We have master routers, defining machine and time required on each machine. As mentioned most product is new and different yet features are similar, just in different configurations. Each configuration could result in different machine than the previous job. Be interesting to see how this works.

    I have put together a schedule using production scheduling, using 1 product. Trying to figure out how to ad multiple routings into it.

Leave a Comment

Scroll to Top