Episode 10

Discover hidden capacities in your Plant

Category :  Manufacturing & Supply Chain

Many plants are often marred by poor efficiencies, late deliveries, and a perpetual fire-fighting mode. After decades of efforts, most of them have given up on solving these chronic problems. However, there is a way out, an easy one, if one can understand the core conflict.

This episode gives is an understanding of the problems, the core reasons for these problems and solutions which can dramatically improve any manufacturing plant's performance within a very short period of time.

Transcript

Introduction: Welcome to yet another episode on the Counterpoint Podcast. We are going to take a step inside the plant today and look at what are the troubles the plant operations go through. What are the core reasons behind it and how we can begin to fix some of them so that we move to a smoother flow and higher productivity in production.

For the discussion, we have Shailesh Ranjan with us. He is the Director at the Vector Consulting Group with an expertise in implementing TOC solutions for all kinds of operations.

Welcome Shailesh.

Shubham Agarwal : Shailesh some time back I happen to visit a plant which manufactures modular kitchen for the residential segment. I saw a peculiar sight right outside the Finished Goods area, I found it very strange to see lot of trucks waiting. I was curious to know the reason, so I asked the Plant Head, whom I knew fortunately.
Shailesh Ranjan : What did he tell you?
Shubham Agarwal : Well he said what you are seeing doesn’t happen every day but mostly around month-end and the reason is that lot of orders get completed closer to month-end, and therefore, the dispatches are skewed.
While he explained the phenomenon to me, I got thinking and wondered that it cannot be so that many customers require a kitchen at month-end! It must be something inside the plant that is happening. Infact we have covered the Month-end skew phenomenon on the podcast, however that was from a Sales perspective. So, is this a one-off case? Do you see this phenomenon in other Make to Order companies?
Shailesh Ranjan : This is definitely not a one-off case. The problem of month-end skew is quite chronic in many MTO companies across different industries. In plants where there is assembly for e.g. kitchen/office furniture, equipment manufacturing, one would see a production skew towards month-end due to better synchronization of components required for assembly in the last week of the month. This leads to dispatch skew at month-end.

Similarly, if you look at plants such as textiles/garment manufacturing, leather processing, where an order has multiple items and has to be delivered together (esp exports), you may not see a production skew but the order completion improves towards month-end. This leads to dispatch skew.

And the funny part is some of these companies claim very high on time delivery!

Shubham Agarwal : You said the order completion improves towards the month-end. It is as if the components required suddenly appeared towards the end. How is that possible?
Shailesh Ranjan : Absolutely.

They don’t appear suddenly. Let me explain with an e.g. suppose a product requires 10 different components. Even if one component is missing the product cannot be completed. And the plant is manufacturing not one but different products and each product requiring multiple components. The different WCs in the plant are busy making components but whether that is helping in product completion, that is a question mark. What happens at month-end is expediting of missing components required for completing the product and that is why you would see better synchronization of components required for product completion.

But why this happens, just hold onto this for a moment. …………………..

Shubham Agarwal : Got it but I don’t understand the connection between month-end skew and companies claiming high on time delivery? Can you elaborate?
Shailesh Ranjan : Sometime back you said how can it be that many customers require kitchen at month-end or for that matter any product! Do customers require their order at month-end?
Shubham Agarwal : Probably not. Orders can be required anytime during the month!
Shailesh Ranjan : Correct. Importantly the required dates would be random throughout the month. There cannot be a pattern which is repeating every month.
Shubham Agarwal : Correct. But I still don’t understand your statement that companies having month-end skew, how can they have high on time delivery performance?
Shailesh Ranjan : If the customer required dates are random through the month and there is a production/dispatch skew at month-end, can the plant claim high on time delivery performance?
Shubham Agarwal : Can’t be. But in-spite of having a month-end skew, can companies really show high on time performance?
Shailesh Ranjan : Well people find ways to hide the problem. For e.g. even if an order is completed anytime during the month rather on the initially committed customer date, it is considered on time. Similarly if an order has multiple items and have to be delivered together, if the order is delivered partially it is considered to be delivered on time even though it means nothing to the customer.
This way some plants show high on time delivery performance but the reality is different.
Shubham Agarwal : I feel jittery about the fact that based on one symptom such as month-end skew, you are claiming poor on time performance. What are the other symptoms which would corroborate poor on time performance?.
Shailesh Ranjan : If a plant is highly reliable with on time delivery in the high 90s, would a plant experience customer urgencies or would customer follow up for their orders?.
Shubham Agarwal : Unlikely.
Shailesh Ranjan : Correct. In such a case there wont be any follow-up calls/urgencies from customer
to expedite the order.
Also if a company is quoting a delivery period as 3 to 4 weeks or 4 to 6 weeks, as a range, what does it say about the reliability of its delivery?
Shubham Agarwal : I wouldn’t say the reliability is very high. Customers in fact increasingly expect a specific date these days for the delivery? Isn’t it?
Shailesh Ranjan : Exactly. If the plant is highly reliable then the quotation period should be a specific date as opposed to a range such as 3 to 4 weeks. The range reflects the fact that you are not sure about the delivery period!
Shubham Agarwal : I am sure that poor on time delivery is not a one-off case and probably affecting many companies. But with so many technological advances such as ERP, APO, robotics/automation, cutting edge machines have been deployed in manufacturing companies (esp. in Make to Order environments) in the last couple of years. Haven’t these initiatives helped them in terms of delivering orders on time or reducing their order execution time?
Shailesh Ranjan : Well unfortunately No. Otherwise we would be out of business…laughs.
The trouble is that they have not solved the problem at the root. They are doing symptom treatment. For e.g. While trying to expedite a late order, if the machine breaks down at the last moment or the operator is not available, one thinks that the “real” problem is machine reliability or absenteeism. It’s similar to I left late from home & when I miss my flight, I blame the traffic.
Then initiatives such as having better machines, deploying TPM, automation, IT for better visibility, etc are undertaken to address the symptoms. But if one looks at on time delivery or order execution lead time, you wont be surprised to see that the performance is more or less the same as before. Reason being that the root cause or the underlying conflict has not been solved.
Shubham Agarwal : Hmm…. So you are saying unless the root cause is addressed, the plant performance will not improve?
Shailesh Ranjan : Absolutely
Shubham Agarwal : Then can you elaborate what is the root cause in this environment?
Shailesh Ranjan : In the endeavor to utilize all the work centers to the maximum with minimal change overs, the planner releases a plan (say monthly) with a loading usually more than the current capability. The thinking is that if you load more, it will create pressure and this will lead to better utilization and higher output. However, this is not true. When one loads the plant more, it provides a large choice set of orders to every work center. By large choice set I mean it would be a couple of days load at the WC. As a result every WC “cherry picks” (products that give better productivity) or clubs orders to form bigger batches (due dates of each order could be far apart) in order to minimize set-ups. Moreover, the cherry picking or clubbing considerations are different at each WC.
Shubham Agarwal : But why do they cherry-pick? I am sure it is not just for fun. What is a need that gets fulfilled when WCs cherry-pick?
Shailesh Ranjan : They cherry pick for better productivity. Remember each WC is being measured on productivity/utilization! Each WC would take steps to look good on these performance parameters. But whether this helps the system as a whole, we will see…
……….
As a result, what happens is – if an order had a voice and whenever its being worked upon it would say “I am working”. What do you think the order would say most of time till it is complete?
Shubham Agarwal : Not sure. It would only be worked intermittently. Right?
Shailesh Ranjan : Correct. Most of the time the order would be mum! Meaning it is lying idle, though the machines are not idle because they are working on something else. So, an order waits lot of times at different WCs before it is complete.

Therefore, the practice of loading the plant more, increases the WIP and elongates production lead times. As the month progresses, the pressure of billing and customer follow-up/urgencies forces the plant to focus on order completion. This is the time when different components required for an order or product start coming together, resulting in production skew towards the month-end and poor on time performance.

Due to poor plant reliability and a production skew at month-end, Planner can make the next month’s plan only after month-end, considering back-log and next month orders. So by the time the next month plan is “frozen”, the plant undergoes starvation in the beginning of the month. This starvation along with urgencies (esp towards later part of the month) leads to loss of output and this in turn puts further pressure to load more. The plant is caught in a never-ending vicious loop of poor on time delivery, loss of output, higher order lead times and high inventories.

Even though intuitively the Plant knows that if they load with few orders they would get better on time delivery and lower lead times. But the fear of poor utilization, forces them to load more.

Every plant faces this conflict whether to load the plant with more orders (for better utilization) vis-a-vis loading with fewer order (for faster order completion). Companies try to resolve this conflict by monthly planning, but as explained earlier the lead time expands, lot of production happens towards month-end, on time performance goes for a toss and plant loses capacity. In a nutshell, the conflict remains unresolved. This conflict is at the root.

Shubham Agarwal : Hmmm…I understand that the plant people have not been to solve this conflict. But what about academicians. Haven’t they tried to solve this?
Shailesh Ranjan : Well they (including IT folks) tried to solve the conflict by deploying a tool which would take into account precise capacities of each WC, loading the plant according to the capacities defined and could also quickly do corrections in the plan when things go haywire, rather then waiting till the month-end. I am talking about APO.
Shubham Agarwal : Did APO or Advaced Planning & Optimization resolve the conflict?
Shailesh Ranjan : Not at all. Firstly, getting precise capacities itself is a big challenge due to product mix impact. Plants handling many varieties with varying process times, this can be a nightmare. Therefore, loading the plant exactly based on capacities defined itself is a question mark.
Even though the plan could be re-scheduled easily & often, but the practice of releasing a plan for the month continued. Due to this practice, cherry picking/clubbing future orders also continued leading to all the symptoms as explained earlier. If the plant is highly unreliable, the ability to re-schedule quickly is of no use because no Plant can follow a plan which keeps changing often!
Shubham Agarwal : So how do we get out of this mess? How do we resolve this conflict?
Shailesh Ranjan : Before I ans this. Let me give you an analogy. Particularly on Monday morning or Friday evenings, what do you see at the airport?
Shubham Agarwal : Lot of rush…
Shailesh Ranjan : True. One would see long queues at the check-in counters. There are 2 types of people one would see in the queue – people like us who get to the airport at the last moment (have a flight in the next 45 mins to an hr) & others who have a flight in the next 2-3 hrs (might have come thru an international carrier & has an onward connection). Suppose you cannot add more check-in counters, what problem would you see?
Shubham Agarwal : Some people may miss their flights…
Shailesh Ranjan : Correct. So you would see two things. 1) Some people missing their flights and 2) Some people check-in too early.
How would you solve this problem?
Shubham Agarwal : Not sure…
Shailesh Ranjan : Well do not allow people to check-in if their flights are departing after say 1.5 hour.
Now can you apply this learning to the plant.
Shubham Agarwal : In the airport example I can understand people missing their flights equivalent to late orders, but you didn’t talk about orders completing early in the plant?
Shailesh Ranjan : That’s a good question. Earliness in the plant you would see in the WIP i.e. orders where due date is much later are also being worked upon. Also in case of a basket order, one may see some line items being ready but not the complete order.
Shubham Agarwal : Ok. Now I understand
Shailesh Ranjan : Great. Now can you apply the learning from the airport example to the plant.
Shubham Agarwal : May be allow few orders at a time…
Shailesh Ranjan : That’s gud. You are getting there…
Shubham Agarwal : But earlier you talked about the other side of the conflict that if we allow the plant with few orders then there is a fear of poor utilization of resources. So how is this taken care of?
Shailesh Ranjan : Let me first explain what I mean by allowing few orders and then I will explain how utilization fear is addressed.

In most environments where touch time (i.e. just the machine process time & no waiting time), is less than 10% of total production lead time, visibility to the plant can be reduced by half. This means in a plant where production lead time is say about a month, the plant is allowed to work only on those orders which are due in the next 15 days instead of allowing all orders required in the entire month.

Since the choice set available at each WC is limited (around half from the earlier case), it prevents “cherry picking”/clubbing of future orders for bigger batches. This leads to faster order completion.

Now coming to the question if we allow few orders in the plant how will we prevent poor utilization of resources.

Way out is to deploy a solution with Push-Pull mechanism. Push mechanism is from planning perspective. Orders are scheduled chock-o’-block on the constraint resource (because CCR determines the output of the plant) based on its current capability. Approximate capacities are good enough. We don’t need a precise definition unlike APO (will explain that a bit later). Order due dates are determined based on the current pending load and not by overloading the plant.

The Pull mechanism is deployed in execution. Since CCR determines the output of the plant a certain WIP is maintained between the CCR & first WC, which is half of the current WIP (as explained earlier). The objective of this load/WIP is to not only prevent cherry picking but also starvation of CCR. So for some reason if this load drops below the desired level, more orders are released into the plant. Similarly if this load is going up due to lower output, further release of orders is stopped.

In a nutshell one is conservative in Planning by not overloading the plant but being aggressive in execution by pulling orders when the load goes down. By doing this the plant becomes highly reliable and the need for re-scheduling the plan goes away. One is able to achieve not only higher CCR utilization but also perfect de-coupling between Planning and Execution which was next to impossible in the previous scenario.

Shubham Agarwal : So are you saying if one deploys the Push-Pull solution, the plant will achieve high on time delivery and reduced order execution time?
Shailesh Ranjan : That is just the starting point which gives good results but the journey is not yet over. A priority mechanism is also put in place which depends on how close the order is to its due date. So an order whose due date is say tomorrow vis-à-vis another one after couple of days, the first order would get a higher priority at the WC. This priority is shown visually at the shop floor which is easy to understand and this entire system runs almost on auto-pilot.

The entire endeavor is to ensure how orders complete in a faster time i.e. reduce production lead time and thereby, facilitating high on time delivery. For this we continuously collect data on where the orders are waiting for the highest time and then launch an improvement initiative to reduce this waiting time. For e.g. if the waiting queue is high before a particular WC, then the improvement project would explore how productivity can be improved so that the waiting time reduces.
For such improvement projects, Lean, six sigma techniques are used to improve productivity of that WC.
These initiatives are not taken once but continuously. Its called POOGI. It not only helps reduce lead time in the plant but also increases output of the system.

Shubham Agarwal : So a plant has to implement all the above solutions to get the desired results?
Shailesh Ranjan : Yes more or less..
Shubham Agarwal : What benefits would one see in such a plant ?
Shailesh Ranjan : WIP and production lead time reduces by half. Fire fighting/urgencies from customer gets eliminated due to reduced lead time and on time delivery is in the high 90s. The flow of material in the plant is faster (since the plant is working on few orders)/no jams, therefore the out put from the plant is staggered throughout the month rather than a skew at month-end.
Also better flow (no starvation at beginning of the month) without urgencies/re-scheduling, the output of the plant also increase significantly (by atleast 30%).
Shubham Agarwal : Wow this sounds exciting.
Shailesh Ranjan : and the beauty is this change does not require any investment or additional resources.
Comments

Your Comment

Your email address will not be published. Required fields are marked *