THE CONCEPTS OF OPTIMIZED PRODUCTION TECHNOLOGY OPT - THE WAY FORWARD M. C. Harrison, Creative Output U. K. Ltd. 16S Introduction Optimized Production Technology, or OPT, is a new concept in the control of manufacturing companies. It embodies both software and a philosophy which challenge the core of some of our current systems and control techniques. It starts from first principles by looking at the goal of the manufacturing organisation, and then shows us how by utilising the large investment already built into our current systems in a better way, we can challenge and beat competition from the East OPT has been proven in a large number of major US. Corporations and is now recognised as a major enhancement to the traditional Material Requirements Planning, or Just.in-Time systems. ‘The Japanese Threat Over the past twenty years, British, European, and ican industry has invested many milions of pounds in t88 implementation of computer based Material Requirements Planning systems. Frankly, the results have been disappointing and largely restricted to improvements created by the mechanisation of previously long-winded manual routines. Ifwe look tothe East, the Japanese on the other handhave performed a manufacturing feat that has left much of our industry gasping or deade.g,, motorcycles, televisions, ‘cameras, and now cars. We wonder how they achieved it, and reassure ourselves that it can be put down to adifferent culture or work ethnic, jobs for le, docile labour and attention to quality ‘An analysis of Japanese and Western companies duting the seventies by Booz Allen and Hamilton shows the extent of the problem. A prime measure for these purposes is Inventory Turnover. An improvement in this figure ‘demonstrates increased throughput or reduced inventories ‘or a combination of both, pointing to better profits, shorter productivn lead tines ond a healthie: company. During the seventies (Fig. 1), Western Industry on average turned its 3.7 times, improving 20% during the decade. The Japanese turned their stocks on average 5.7 times, but ‘more worringly the improvement over the decadewas 31%, Not only were the Japanese better, but moving quickly further ahead. Further information into the 1980's shows that the gap continues to widen. ‘A more detailed analysis by industry type (Fig. 2) shows where the area of greatest concern lies. In both Job Shop ‘and Process industries the difference between the West and. Japan is minimal, But in the Repetitive Batch manufacturing ‘sector, by far the largest, the Japanese turn their stocks. twice as fast as the West. No longer can we point to cultural and other reasons for the difference, we must look in detail ‘at our methods of manufacturing and production control 1S ORE RO NT Fig? ‘WweTOR TURNER By OMFG (USTRY umn a ee a= es a aa ss aa ome pe a FER, Ie - 1579 SSOMTIALY. THE ENTE Ge 1 MANTOY TRON IS I RFE PaOFCTADG, Fo invesony — USA/Europe ssapanl owen Boos Aen & Faron Survey aE Companies What is the Goal? ‘We must start by asking what is our purpose in the field of production control. It must be not just to employ people, to produce product or to improve efficiencies on the shop floor. It is to ensure that the company MAKFS MONFY. But not just profit on its own. In order to survive and improve, a company must simultaneously improve NET PROFIT, RETURN ON INVESTMENT and CASH FLOW. Any system which fails to consider these facts. as its purpose, is destined to drive us in the wrong direction. But these are Financial Measures which one applies to a complete plant or business and which are dificult to ‘monitor on a daily basis, so we need to derive Operational Measures which will help us with our daily decision making. ‘The first can be specified as Throughput (T), not output of finished product to warehouse, but the production of items which will immediately generate cash through sales. The second is Inventory (I), including Raw Materials, Work-in- Progress and Finished Goods, the levels of which reflect the cash tied up and directly, thelengthof production eadtime. The last and ‘catch all’ is Operating Expense (OE). Operating Expense covers all the expenditure of the AICS CONTROL JUNE/JULY 1965 7Company other than that related to the purchase of raw ‘material and components for conversion into Throughput. ‘So what are the aims of the company and its production control system? We should aim to increase Throughput, whilst SIMULTANEOUSLY reducing Inventory and Operating Expense (Fig. 3). To improve one without con- sidering the effect on the other two, may take us in the wrong direction. ‘We will use these criteria to judge the effectiveness of our systems. Fo? GOAL OF MANUFACTURING wot ait ot Operations! ft ft of, i Balance the Flow - Not the Capacity Now that we have set the goals, what do we need to do to achieve them? The Japanese use the analogy of a river system to explain that the allimportent ruleis to balance the flow of production. One has only to look at the process industry or the assembly line to see two examples where material flows smoothly with oniy limited inventory in the system. Both are considered very efficent, and yet everyone will realise that, ‘in the line’, there are processes ‘working at significantly less than their capacity. So why is it that we don't recognise the same phenomena in batch production? It is impossible even in a dedicated assembly line to balance the capacity of all machines exactly. How much more difficult is it to balance capacity in a general purpose- machine shop? We should recognise capacity imbalances and concentrate on achieving fast efficient flow through the resources with a minimum of inventory in the system. This leads us to the first_of the rules of facturing developed by Dr. E. M. Goldratt, known as. ules of OPT. RULE 1; BALANCE FLOW NOT CAPACITY, Understand the Role of the BottleNeck Ifitis impossible to balance capacity, we have to recognise that in any plant there will betwodiferent kinds of resource bottlenecks that constrain production and . non- bottlenecks which have spare capacity. Production control js an attempt to organise our resources and the flow of materials to produce a schedule of operations which leads us towards the goal of the manufacturing organisation. In order to do this we must first understand the inter relationships between the two types of resources we havein allour plants:-bottlenecks and non bottlenecks. A resource ‘can be any element needed to produce a product such as people, equipment, machines, fixtures, quages, space and the like, Let us denote a bottleneck resource as X and assume thet the total of all market demands for this resource is 200 hours per month. Letus further assume that this demand exactly matches the available potential or capacity of this resource (200 hours/month). Let us also denote Y as a noncbottieneck resource with market demands of 150 hours per month and a potential of 200 hours/month. We can now examine the four different telationships between bottleneck and non-bottleneck resources which comprise the fundamental building blocks of any manufacturing operation (Fig. 4). Fig ‘THE BUILDING BLOCKS-OF MANUFACTURING TE] sornencox resource S eemanb kas yon 1 potesnac sone wows ‘= DEMAND = 180 HRS /MONTH 1 POTENTIAL» 209 HRS MONTH gases case Usticn fomm TSH Uaton mS Customer Demancs ye] Oy x Uation 100% 78 © Case I- Allproduct flows from X to. In this situation, we ‘can fully utilise resource X 100%, but we only utilise resource Y 75% of time. X’starves Y. It does not produce enough product to keep Y working all the time. * Case 2 - All product flows from Y to X. Again, we can utilise X 100% of the time and if there are sufficient raw materials we can activate Y 100% of the time. However, since our goal is to simultaneously increase throughput and reduce inventory and operating expenses, we must conclude that we can only utilise Y 75% of the time. To activate ¥ more than 75% of the time results in the build up ‘of workinprocess inventory ahead of X. This action increases inventory and operating expense without increasing throughput. It moves us away from the goal of manufacturing and should not be called utilisation # Case 3 - X and Y, instead of feeding each other, feed a common assembly. Again, we can fully utilise resource X 100%. However, if we attempt to activate resource Y beyond 75%, we will build finished parts inventory in front of assembly. So we must conclude that we can only utilise resource Y 75% of the time. # Case 4 - Kesource X and ¥ do not feed each other or a common assembly, but feed completely independent market demands. Once again, we can fully utilise resource X 100%, but we can only utilise resource ¥ 75% ofthe time. To activate it above this level results in finished goods inventory, since the market demands for ¥ are only 150 hours/month, There are many other relationships between bottleneck and non-boitleneck resources which we could consider. Fortunately, we have obtained exactly the same answer in each of the above situations. Therefore, we can conclude that we will get exactly the same answer for any linear combination of these four cases. Linear combinations of these four building blocks can be used to depict every type of manufacturingand every manufacturingplant that exist. So what we have is not four relationships between bottlenecks and non-bottlenecks, but a representation ofthe reality of manufacturing. Consider the more normal diagram of a product manufacturing process (Fig. 5) showing both the structure and the operational routings. We can choose to put the bottleneck at any operation. But whenever we do so, we find that the relationship between the bottleneck operation and any non-bottleneck operation is governed by the four cases alone. Consequently, we can derive two further rules of OPT, Figs SINCE THE ANSWER IS THE SAME IN ALL FOUR CASES, IT WILL ALSO BE THE SAME IN ANY LINEAR COMBINATION OF THESE FOUR BASIC BUILOING BLocks. production at the bottleneck. It is equivalent to an extra hour of throughput for the TOTAL system. ‘On non-boitleneck operations we have three elements of time processing, set-up, and idle time (Fig. 6). Here, if we save an hour of set-up, we gain an additional hour of idle time. Consequently, an hour saved at a non-bottleneck is worth nothing Fig ‘SET-UP - DEPENDS ON THE TYPE OF RESOURCE [Ez] sorenece jj enocessme 1 sera “+ 4 SET-UP HOUR SAVED RESULTS IN| AN ADDITIONAL OUR OF PROCESSING TIVE + AN ADDITIONAL HOUR CF PROCESSING TIME RESULTS IN-AN ADDITIONAL HOUR OF THROUGHPUT FOR THE TOTAL MANUFACTURING SYSTEM TE] semsorrence|, mnogesne | serur | wie “+ A SET-UP HOUR SAVED RESULTS NAN ADDITIONAL OUR OF 1OLE TIME {+ AN ADDITIONAL HOUR OF IDLE TIME 1S WORTH NOTHING = ITS A MIRAGE RULE 2 - THE UTILISATION (profitable) OF A NON- BOTTLENECK RESOURCE IS NOT DETERMINED BY ITS CAPACITY BUT BY SOME OTHER CONSTRAINT IN THE SYSTEM. RULE 3. ACTIVATION (making it produce) OF A RESOURCE IS NOT THE SAME AS UTILISATION (prof- itable) OF THAT RESOURCE. (Our experience shows that there are very few bottlenecks jin a plant (usually lacs than 3%). We must therefore rise that chasing the goal of high machine usage (GVATION!) will not improve production or throughput but will generate excess inventories instead. ‘Our systems must therefore reflect the need tounderstand the effects of finite capacity constraints on other parts of the plant. We must schedule non-bottlenecks around the capacity constraints, and not ignore them. Before we move on, consider the impact of what we have just said on the traditional cost accounting measurements we use to monitor our plants. They tend to look for hich machine usage and high output of standard hours (irrespective of ‘actual capacity requirements), and often drive us in the ‘wrong direction. But that is the subject of another paper! The Value of Bottleneck Management Traditionally, we regard all our resources with equal or near-equal value, We must learn from the above the difference in value of the two types of resource and how ‘great is the payback from extremely tight control on a very limited number of resources -the bottlenecks. Let us first consider the time available on each type of resource, and how it is divided between different kinds of activity, The available time at a bottleneck resource is split between processing time and set-up time (Fig. 6). Ifwesave an hour of setup, we gain an hour of processing time. Furthermore, an hour saved at a bottleneck operation is vastly more important than just the increased hour of ‘This leads us to two more OPT rules: RULE 4: AN HOUR LOST AT A BOTTLENECK IS AN HOUR LOST FOR THE TOTAL SYSTEM. RULE 5: AN HOUR SAVED AT A NONBOTTLE- NECK IS A MIRAGE, Our schedules should recognise that at, bottleneck operations it is very important to save set-ups. However, at non-bottlenecks, there is no benefit from saving set-ups. In fact, we would benefit from running more set-ups because our lot sizes would be smaller. While these smaller lot sizes would not increase throughput, they would decrease inventory and operating expenses. But further than thel consider the way we control ‘machines now. Much downtime is due to lack of tooling, ‘operator absence, wrong or absent materials or drawings. In most medium sized companies, the value of an hour on the primary bottleneck is between £10,000 and £20,000 in Throughput. Have you checked that on your bottlenecks there is a relief operator standing by, that all materials are available to the machine before they are needed, that your machine spares policy reflects a downtime of £10,000 per hour?! Does your bottleneck stop for lunch?! By simply expending effort to ensure that a very few machines, the bottle-necks, are correctly looked after, we ‘effectively improve the productivity of the whole plant! Priority and Capacity We now understand the importance of capacity constraints, in the production process. But how should we plan production accordingly? Let us first consider how current Wester systems work. ‘Our current scheduling logic says that the item with the longest leadtime should be released or run first. Priority and capacity ere essentially considered sequentially, not simultaneously. Our typical MRP_ scheduling logic generates priorities through leadtime offsets. Once the priorities are established, we check to see ifthereis roughly enough capacity in each time bucket to meet these schedules. However, we do not check the interaction of Priority and capacity within or across the boundaries of BCS CONTROL JUNE/JULY 1995 11