What is APS and why it doesn’t “make a production plan the way we want ...”

Features of commercial APS algorithms

Peterkin S.V.



"Advanced" commercial APS usually consists of 2 parts. The first conditionally can be called APS – synchronizers, their task is to calculate (synchronized and optimized) plans for the entire "tree" of the product (hereinafter - SI - Product Composition) or order, and / or for the entire production-supply-supply chain. The second is APS-schedulers, the task of which is optimization according to some criterion of passing production, unrelated tasks through a group of equipment. The former, as a rule, when planning, consider all resources (materials, production resources, ...), the latter, as a rule (there are exceptions), - only production. APS-schedulers under different names act as "schedulers" in MES-systems.

APS – synchronizer, “classic” approach

What is the essence of APS-synchronizers? These are powerful “calculators” that can, with certain assumptions, quickly draw up a detailed and accurate (at the time of calculation!) Production plan. In many cases - and with optimization elements according to various criteria. This is their big plus and the possibility of their easy sale: “here it is, that very magic button that will consider me a real (really feasible) plan”! But will it be? ..



Let's take the next advantages of APS: it doesn’t plan into the past, it plans taking into account resource loading (people, cars, tools, ...), each time it creates a feasible plan, etc.

And now the "dark side of power." And above all: yes, plans, but at what cost? Price:

1. the presence of an accurate operational process technology and an exact resource (people, cars, ...);
2. organization of accurate and operational accounting (accuracy - no worse than “before the shift”);
3. powerful hardware platform;
4. price (money and labor for administration) ...
5. working “adult” (based on MRP-II) general planning system / methodology (SOP -> MPS (DP) -> “MRP” (APS) -> “Scheduler”)

We rarely pay attention to the last condition (and actually, few even suspect that there is no “in the nature” “production planning” consisting of one (super) plan ... And this condition is a defining and: non-decision upper level planning tasks leads to GIGO (Garbage-In-Garbage-Out).



Further. The disadvantages of the algorithm when used in real conditions:

1. in case of non-compliance with paragraph 1 above - an inaccurate plan that does not fall into the "launch window" - workers are standing ...
2. if item 2 is not observed above, the plan shifts to the right. And so - every day ...
3. multi-product manufacturing of complex products and problems according to claim 3. Above is a great system response time.

p5 - omit: before you run, you must learn to walk ...



Plus, more interesting APS features:

1. excessive complexity and "heaviness" of the algorithms, multiple overkill for most applied "our" production tasks, and at the same time, almost complete impossibility to solve the following, "unimportant" problems;
2. "Sharpened" APS on the western industrial model, the model of an absolutely mature and absolutely market factory, and the "western" dimension;
3. the inability of APS to rigidly reserve stocks and production tasks (expected arrivals) for any head order. With each re-planning, APS will “take away” reserves of both capacity and stocks for orders that have become more priority. Hence the extremely “nervous”, ever-changing plan. Moreover, in complex mechanical engineering / instrument engineering, in production they like (and this is really necessary, why - a separate conversation) to rigidly reserve the inventories and materials for specific overhead products (under the serial number of the manufactured “machine”). But APS does not want to know anything about it;
4. Despite the fact that most discrete APS are custom, with each re-planning, they “forget” about the plans of the last order, and build a new one. In tough order production, where tough order bundles are needed from the parent product to the material, this is not applicable;
5. APS build Mega-accurate and optimized plans. Every time - new. Each time - with the "extreme", shifted to the right, a new date for the execution of the order. And if nothing is already started, they will even show deviations of the original plan from the new one. But, if a lot has already been launched, i.e. there are production tasks, and among them there are lagging ones, APS will hardly react to this in any way - it will give a signal to launch a new one or shift the entire order network. Not showing which and how many PPs are behind. For our "traditionally oriented" and even departing from "traditional" industries, which tend to fulfill orders at the end of the period, this is unacceptable;
6. there are also features, for example: it is extremely difficult, if at all possible, to implement “TOC MTA” management with APS, with a rolling calculation of the “forecast” of achieving the “red” level of general, “WIP + finished goods” stocks;
7. quoting close to Goldratt’s text (“Necessary, but not enough”): because of their rigidity in accuracy and planning detail, APS “coolly” spread local “disturbances” to the entire production, the entire “order network” ...
8. ... and one more short remark: in commercial APS there is no "binding", i.e. A transactional system designed to collect information and stocks, the course of production, to automate and manage (electronic) production workflows and objects (probably the only "classic" APS form of "simple" analysis is the Gantt chart. Many "near-workmen" are admired in life it’s practically not applicable). In the West, integration of APS with ERP systems is implied (and, attention, repetition! Mandatory logical integration with the general system of plans), but we practically do not have any of them in production (correctly implemented and correctly used). And the very idea of ​​the applicability of an outdated ERP model now is very doubtful.

Setting up a system of “collecting” and “processing” information from scratch is at least half the total effort to build the right production system.



Are APS completely hopeless? No, just like ERP. With the condition that you approach APS as a “simple” calculator, or, better, an almost direct analogy, as a processor. If you need to build a computer, under certain conditions, you need a processor. But there are many, many, many other things ... that same “binding”, item 7 above. And at the same time - be prepared - it will be necessary to "slightly get into" the processor, to correct some features of the algorithms for you ... And this, in the case of the processor and even a very skilled assembler, is almost impossible. With APS - it is theoretically possible, practically - ... But, if you know how to do it (and collect and “file”), then - good luck (and this is no laughing matter). If not, get yourself a Procrustean bed APS. And it will be worse than the "cold and bony hand of ERP" "out of the box" ...



If an alternative? For "our" industries?

We created on the basis of 20 years of practice in building a methodology and implementing IT systems, including (long ago it was ...) those same ERP and APS. How is our approach different? Simplicity ... Having realized almost all the capabilities of the “network” APS (network custom planning, capacity utilization (according to the resource model), and some others) and eliminating the above disadvantages, we, in SCMo, build the concept of planning on the simultaneous existence of two types of plans: "directive" ("ideal" "just in time" plan) and "settlement".



Which are constantly compared with each other, taking into account the situation in production, including, with running production tasks. Thus, giving production managers what they most lack: not a super-precise plan (maybe it (!) Is needed for a short period of time ...), but deviations are bottlenecks both in production and in the “network” »Order. The correct definition of the "right" production plan: "The plan does not have to be accurate, it has to be reliable!" And the worker. What else is special with us? The interface part (on-screen monitoring forms. In the system itself — not BI) and the obligatory “fitting” (sometimes with the “kernel” fitting) of calculation algorithms to the particular features of the enterprise’s parallel production system (we will never automate plants according to the employees ...) Higher, in comparison with equivalent APS, calculation speed, etc.



What have we lost with this approach? Nothing. (mega) Optimization, i.e. drawing up a plan taking into account all and all sorts of different, different restrictions - is not used at the beginning of the system according to our methodology. But not because the system cannot, but because production is not ready. Is it tangible? So far, and in the short term - no, because all these “little things” are the essence of the internal workshop management. And then - why should we “catch fleas” in the “operational planning of the machine park” if the production of the workshop / site is poorly synchronized with the production of the other “partners”? (this, by the way, is a typical approach of many "pit-sellers" of MES: MES is optimization of internal shop work, without regard to the others). And if the plans for the production chain are synchronized, isn’t it easier to manage the “inside shop” of the Lean toolkit?



And if, nevertheless, someone wants to optimize already at the level of inter-shop / district planning? No questions! Moreover, within the framework of the same model (it will be necessary to detail the resources ...)

APS– scheduler

They are the “calculators" of MES planning. Their place is intra-shop / district planning. It is possible and justified if and only if the production chain is synchronized. And if “seriously” - then and only if p5 above is implemented (a general system of plans is built). In this case, in the presence of clearly defined and difficultly planned manually bottlenecks, they can be applied. First of all, to determine the sequence in the processing queue in front of a specific site / machine or successive groups of equipment. Because of this, APS – schedulers “started their life” as sequencers, as a rule, in simulation systems.



If you have custom-made production, each order is significantly different in terms of equipment setup and processing, orders / processing and processing options - a lot, you can’t pre-set the rules for their placement - yes, in this case this part of APS can help you. Considering that you are ready to pay the price indicated above (not all “features” of APS – schedulers are indicated). And given the following comment: if the "muda" generated by this part of the APS is less than the "muda" without using it. The first muda is the inevitable normative increase in the execution cycles of the order, necessary for setting the “optimization window”, in which the scheduler will “shuffle” production tasks at its discretion, and the associated growth in wages; the second - losses on changeovers, loss of bandwidth bottleneck, the risk of failure to meet the deadlines for the execution of the order. In my practice, in discrete manufacturing, it is extremely rare that an APS – scheduler is needed. Many tasks of intra-shop management are solved by “primitive” Lean and / or TOC tools, as well as “simple”, through IT methods, by visualizing what happens with semi-automated queue management elements.

"Continuation"

The use of the “correct” APS / MES in the “continuous” (“where everything flows”), metallurgy, food processing, and for solving some other narrowly specific tasks in production where equipment (its use) is critical for the process is not discussed. They (systems) simply have to be there.

Abbreviations

SCM, SCMo: Supply Chain Management and Supply Chain (Planning) and Monitoring - production chain management, planning and monitoring of production chains. A set of methods and automated systems based on the classic MRP-II concept that allows you to plan production and procurement synchronously, taking into account existing and future material and production resources. Both within a separate enterprise and for distributed systems (production and supply chains).



APS - Advanced Planning and Scheduling , Advanced Planning, or, closer to the point, synchronous planning and optimization. A class of information systems - production and inventory planners, in which calculating production plans and / or inventory levels using optimization algorithms. In the classical system of plans, enterprises can correspond (depending on the specific system of a particular manufacturer) to the planning levels from DP to Scheduling.



MES - Manufacturing Execution System , production management system (execution!). A class of information systems designed for detailed planning and production management, usually at the level of equipment and operations. In some cases, using integration with equipment. Fundamentally different for different types of production.



ERP - Enterprise Resource Planning - enterprise resource planning. A marketing term and a broadly defined class of information systems that allow integrated automation of certain areas of an enterprise’s activity, as a rule, without planning, budgeting, product composition and some others.



MRP-II - Manufacturing Resource Planning , Manufacturing Resource Planning - the western 1970s of the last century, the concept of production and operations management. In which production and inventory planning is carried out according to MRP (Material Requirements Planning) and CRP (Capacity Requirements Planning) algorithms.