So net-net the dip in profit can entirely be attributed to the operations.
- Labour Hours Distribution Analysis
The Direct Labour – Effort distribution is shown below for the various steps of the manufacturing process for the month of September.
The analysis of the Direct Labour Effort distribution reveals that:-
- Majority of the 5761 boards processed in September go through all the operations. There are a few operations which are not required by design / not carried out by error.
- The Set-up time (one off per order) is significant for the automatic machines like CNC Drill and CNC Router, compared to the Run time (variable per unit of output).
- The maximum amount of Effort is consumed by the Drilling Department. Profiling is the second highest contributor. By optimizing these 2 departments, and making Productivity gains will have the highest impact on the over-all productivity.
- CNC Drill and CNC Router – Optimization
The CNC Drill and CNC Router are alternative automated solutions to the Manual Drill and the Punch Press respectively. These automated solutions are characterized by high fixed overhead and low variable overhead. The fixed overhead is primarily contributed by the computer controlled programming that is needed to execute the job.
Donner has an operations policy today that, CNC drill will be used for orders with order size 100 and above only. This number of 100 appears to be arbitrary in nature.
So the following are the key optimizations that need to be arrived at :-
- The Order-size at which CNC Drill becomes economical over Manual Drill.
- The Order size at which CNC Router becomes economical over Punch Press.
By solving the algebraic equations of the Fixed Overheads and Variable overheads, the mathematical solution tell us that the order size (point of indifference) is 5.9 for Drilling and 200 for Profiling.
This is shown as a simple tabulation below.
At an order size of 6, the Total time for Manual Drill is 255 Labour minutes, and that of CNC Drill is only 252 minutes. At an order size of 200, the Total time for the Punch Press and CNC Router are both 250 minutes.
Graphically the variation of the Total Effort (Labour minutes) for these 2 options are shown below. The intersection point is the point of indifference.
So the Organization can have a revised Operations Policy for Optimization that:-
- For Order size from 6 onwards only CNC Drill is to be used.
- For Order size from 200 onwards only CNC router is to be used.
But please note that currently Altemeyer appears to be the only person doing all the Computer Control Programming for the CNC machines. If this revised organizational policy is implemented then he may become a bottleneck for the process. Hence it is recommended that the other shop-floor team members / supervisors are trained to operate the CNC machine and can create the programs on their own – without any bottleneck / delays.
- Order Size Analysis
The cost of a unit of output is usually driven by the order size. This is especially true in industries where ‘scale’ plays a key role in determining the cost per unit of output.
The interesting combination of the fixed overheads (one off setup costs) and variable overheads (run time costs per unit) give this advantage of ‘scale’.
From the Table above the following inferences can be made:-
- As the Order size increases, the Average Labour Time per Board for Standard Operations reduces. This is the advantage of ‘scale’.
- The optimization of CNC Drill and CNC Router usage helps in achieving added productivity gain when the order size goes beyond the respective point of indifference.
- The standard labour time for select order sizes are :-
-
Order size of 1 boards ➔ 383.4 Labour minutes per board
-
Order size of 8 boards ➔ 76.9 Labour minutes per board
-
Order size of 20 boards ➔ 34.6 Labour minutes per board
-
Order size of 100 boards ➔ 12.0 Labour minutes per board
-
Order size of 200 boards ➔ 9.2 Labour minutes per board
-
Order size of 250 boards ➔ 8.5 Labour minutes per board
- Please note, the times shown above are assuming optimum usage of CNC drill and CNC router.
The reduction in Standard Production time with order size increase is shown in the graph below.
- Capacity Analysis
Capacity Analysis is useful to find out the maximum possible throughput within each process step, considering the resources available at that process step. Hence when there are multiple process steps involved in a manufacturing process, this analysis is useful to determine:-
- The Theoretically possible throughput capacity of each of the process steps considering the resources available.
- The Theoretically possible throughput capacity of the over-all system – which is the lowest of the through-put capacities of the individual constituent process steps.
- The bottleneck – or the process step which has the minimum system throughput capacity – which plays a limiting role on the throughput rate of the over-all system.
The key resource in constraint is the Labour availability. Donner employs 18 labour staff on the shop floor. Additionally the supervisor staff spent 90% of their staff as direct labour. Only the remaining 10% of the time they work as supervisors (in manufacturing overhead).
So Total Labour Capacity available at our disposal is = 18 + (90% * 4) * 7 hr working day * 60 = 9072 minutes.
Given this capacity constraint, the analysis of the Donner process for various Order sizes is shown below.
The following are the key inferences that can be made from the table shown above:-
- As the order size increases, the System Throughput capacity of number of produced per day increases. This is explained by the productivity gain of the reduction in the standard production time to complete a single board.
- The Max possible System Throughput rate for the Order size 10 (The median value of the order size as per the profile of orders received in September) is 349 boards per day.
-
The bottleneck of the overall process flow is Drilling. This has got the lowest throughput considering the available constraint in labour capacity.
-
This bottleneck is true for both manual drilling and CNC drilling – irrespective of whether the order size is below or beyond the tip-over threshold.
The Increase in System Throughput Capacity over Order size is graphically shown below.
As an example, the capacity analysis of the Dry Film Photo Resist (DFPR) area shows:-
- The Throughput of the Over-all DFPR area is driven by the bottleneck in the Laminate and Expose Step. This has got the lowest possible throughput capacity.
-
As the order size changes from 1 to 10, the DFPR area throughput increases from 448 boards per day, to 4032 boards per day.
The increase is DFPR area throughput capacity with order size is graphically shown below.
- September - Order Profile Analysis
Let us focus on September, the month that saw a serious reduction in Profit because of the increase in the Direct, Indirect Labour and the Manufacturing Overhead as proportions of sales.
The profile of the Order Sizes of the orders received in September is shown below.
The Key inferences that can be made from this Table are:-
-
The Weighted Average order size of the 60 Total orders and 5761 PCBs ordered in September is 96.01 boards per order. However this number is not very representative of this population order profile because it lies beyond the 80% probability distribution mass.
-
A better representative number of the boards per order value is the Median of this distribution - which is 10 boards per order.
-
The company policy says that the CNC Drill will be used from Order size of 100 upwards. However by looking at the numbers, it shows that the CNC Drill has been used only from order size of 140. This explains some of the added productivity loss during the month of September.
-
The CNC Router can be used from order size of 200. The data reveals that the CNC router has been used from order size of 229 onwards. This has however not resulted in any productivity loss – because there were no orders between these numbers and 200 happen to be the point of indifference.
- In September, the actual Labour hours spent on the CNC Drill and Manual Drill are:-
-
Manual Drill ➔ 636. 8 Labour Hours
-
CNC Drill ➔ 196.8 Labour Hours
-
Total ➔ 833.6 Labour Hours.
- If they had optimally use CNC drill and Manual drill the numbers would have been:-
-
Manual Drill ➔ 49.58 Labour Hours
-
CNC Drill ➔ 329.87 Labour Hours
-
Total ➔ 379.45 Labour Hours
-
This would have resulted in a productivity gain of 54% in the Drilling Process step.
-
Similarly CNC Router usage would have yielded a Productivity gain of 8.4%.
- September - Load Analysis
Let us analyze the Load that was processed through the system in September and understand where the Productivity loss (if any) happened and the areas where gains can be made.
The Key inferences from the Table below are:-
- 60 orders went through the system in September, Total 5761 boards.
- The median order size of the profile of the orders is 10.
-
If the CNC Drill and CNC Router were used optimally (To-BE process) 30% reduction in standard labour time could have been achieved.
- The key data points we can observe are:-
- Average Effort per Order is 25.53 Labour Hours for AS-IS process. For To-BE process the equivalent Labour Hours could have been 17.96 Labour Hours.
- Average Effort per PCB in 0.265879 Labour Hours. For TO-BE process the equivalent Labour Hours could have been 0.325257 Labour Hours.
The following are the key productivity losses that happened to Direct Labour during the month of September. The estimate of the losses is made based on certain assumptions.
-
Internal QC Check led to identification of defects. These defective orders needed rework and this rework effort has not been accounted for. The reject rate was 1% total rejects and 6% partial rejects.
- Assuming Partial reject needing about 30% effort to complete and full reject needing complete rework, an estimate on the effort lost can be made.
-
Field rejections of the orders made in August were received in September. They had to be reworked and delivered back quickly to the customer. This reject rate was 10%.
- Using the Ratio of the Material Cost US$ ratios between August and September we can estimate the number of boards delivered in August.
-
When Small Rush orders arrive, Arthur Dief personally took care of these orders and fast-tracked these orders to completion within the short delivery times. Because of this prioritization, several other staff working on jobs had to be moved / kept idle.
- The Number of orders was 4 per week, or 12 in total in September.
-
Customer Engineers telephoned he Labour staff and they had to spend time talking to them. Roughly 1/4th of the jobs were delayed. Assuming 5 minutes per phone call, the labour hours lost on telephone can be estimated.
-
Movement between machines by walking across the factory layout, Idle-time caused due to bottle-necks in the system, prioritization, deprioritization issues and poor scheduling contributed to the balance.
Based on this, the AS-IS and TO-BE models are calculated as follows:-
From the Above Table the following key inferences can be made:-
- By following the September order profile – even if rest of the overhead losses – was roughly the same (worst case), just by optimally using the CNC machines, a productivity gain equalling to freeing up 20% of the available capacity can be made.
- This 20% spare capacity can be used to handle ‘extra load’ that is projected to be received in the subsequent months.
- September - Inventory & Flow-time Analysis
Let us look at the Sales Volumes between August and September. The future projected Sales Volume (if flow time can be improved) is 3M US$ annually or 250K US$ per month.
Now, the Opening Stock Inventory on Sep 1st can be approximated to the Average Inventory that would have prevailed in August. The deliveries in August ran with a average day of 10 days. Among this, 4 days are caused by upstream processes (in procurement, order despatch to manufacturing shop etc.). So balance 6 days is the delay in manufacturing process. So Total Flow time of an August order on an average would have been 15 + 6 = 21 days. So, net-net August Inventory works out to 307 boards/day * 21 = 6441 boards.
With this opening stock known and the incoming rate in September (assume uniform inflow) we can calculate the Average Inventory of September using the Daily Shipment Values.
Please note the Daily Shipment Counts include the returned orders (negative values) and hence they have to be handled as inflows into the system for inventory calculation.
It is mentioned that, there are some orders which are Pending Customer Clarification and they are typically on an average of 5 per week, or 5 at any point in time. The cycle-time for these issues to be open is an average of 5 days. So the total inventory of such orders is 250. We need to make this adjustment
So Net-Net the Average Flow-time for September calculates to 27.67 days.
The Key inferences from the above table are:-
- The manufacturing process flow-time for September order was 27.67 days. Adding the 4 days lost in upstream process. The average could have been more than 31 days.
-
To serve the future run-rate of 3M US$ annual revenue, or 250,000 US$ per month :-
-
If the AS-IS process of September was followed, then there is no hope of scaling up to catch up with the demand. The demand would be 240 Labour Hours per month, however even if the 8 people were trained and ready – they wont be able to meet the demand of 15 days flow-time.
-
If the TO-BE process (of optimal use of CNC machines) was followed, then even with the current level of effort wastage overheads (quality issues rework, idle-time, telephone interruptions etc.) we can handle with 4 people.
-
If the Quality issues were reduced and the effort demand for rework by 40%, then none of the 8 people are required – even to service the future demand.
- Summary of Problems Faced
Although we have touched upon it, in the previous sections, the summary of the problems faced are collated below.
-
The Profit margins have dropped significantly consequent to the increase in the contribution of direct labour (lower productivity) and manufacturing overhead (increased supervision, abnormal idle-time etc.).
-
The policy to determine when to usage of the CNC machines. Currently policy says anything more than 100 boards per order only CNC drill will be used.
-
The policy to take more and more small input orders – because they have been successfully delivering small orders with good quality and good flow-time.
-
The policy to quote 3 weeks as order delivery time for orders < 1000 boards and 5 weeks for large orders > 1000 boards. These numbers seem to be arbitrary in nature.
-
Not optimized for either Job-shop or Flow-Shop layouts. It is somewhere in the middle and not getting advantages of either of these layouts. So people ended up walking a lot of distance to move to the next machine / next job.
- The company does not have comprehensive productivity tracking and measurement policy.
- There is a significant productivity loss with inappropriate usage of machines for the context.
- There is a significant productivity loss with productive labour spending more than 52% of their time on non-productive tasks including rework, telephone interruptions, deprioritization of jobs because of a small rush order squeezing in, and walking-time, idle-time caused due to bottlenecks and improper layout.
- There is a lack of consistency in the Quality. On one-hand the small orders are getting delivered within time and also in good quality with zero reject rate. However regular orders are taking almost twice as longer and still having field rejection rate as high as 10%.
- The internal rejection rates are also high of 1% full rework cases and 6% partial reworks cases. This % is lower than the field reject rate which is also an area of concern.
- Delivery time quoted for regular orders was 3 weeks and large orders > 1000 boards was 5 weeks. However orders were delayed significantly and the delay was around 10 days in August, 8 days in September, and 9 days in first half of October. This leads to uncertainty and unpredictability in delivery time.
-
The workload on the staff seems to be poorly balanced across the month. They have very light load in the first half of the month and a heavy load in the 2nd half of the month. This seems to be an effect of poor job scheduling.
- The standard raw material they use appears to be the double sided copper clad glass epoxy sheet. However in spite of this, they seem to be doing supplier search and filtration for every single order and buying at cheapest price. This is leading to a 2 day delay in placing the order – and at-least 2 days delay for the goods to be received into the factory.
- They do not seem to have a clear policy on inventory /stock of raw materials.
- After the order is received into the manufacturing they seem to take 3 to 4 days before the job is scheduled. They appear to wait for the raw material to be delivered before they scheduled the job.
-
The job process scheduling seems to be imbalanced across the month. The first half seems to be leniently loaded compared to the 2nd half.
- The workers were arbitrarily moved to the bottleneck points at will. But there were times when labour staffs were busy doing non-critical tasks - when critical jobs were waiting.
- Small Rush Orders appear to bulldoze through all existing jobs by gaining high priority. The exact impact of scheduling small rush orders as higher priority ones is not well understood by the company.
- They are not clear on the bottleneck points and the bottleneck in Drilling. Although majority of their staff work in Drilling, they do not seem to follow even their organization policy – however poorly it has been defined.
- The racks are buffers for holding the panels. There are only 20 racks which can take 8 board panels. So a sleeping inventory of only 160 units can exist in the system. They do not seem to be having this under control.
- The Average Cost per Board based on the order size, and the Average Delivery-time based on the order size where not available for quoting in the bids.
- The sales potential existed for next year up to 3M US$ i.e. 250K US$ per month on an average. But their operations were not ready to scale to this demand – because to win new business needs strict 15 day turn-around time.
- Recommendations
Based on the detailed analysis carried out in detail in the previous sections of the document, the following recommendations can be made to resolve the problems described in Section-11.
- Organizational Policy
- As outlined in the CNC machine usage optimization analysis the recommendation is to revise the organizational policy to :-
-
Order Size >= 6 orders ➔ Use CNC Drilling machine (else) manual drill.
-
Order Size >= 200 orders ➔ Use CNC Router (else) punch press.
- The organization seems to have defined a policy to take more & more small orders because they are able to schedule & get them through. This policy is not correct and they need to take a good balance of large orders as well to get the advantages of scale. They can stick to the September profile of Orders and don’t need to go solely for small orders.
- By optimally using resources they can easily achieve 3 weeks as delivery time – even with a sales target of 3M US$ as proved mathematically in Section-10. Hence No change needed to this policy.
- The policy to say any order with order-size < 8 as small order can be left as is. Even though 6 orders seem to be the logical point, because a panel can hold 8 orders and the rest of the panel may not be useful, it makes sense to stick order sizes up to 8 as Small orders.
- Production Layout (Job Shop and Flow Shop)
- The company should divide itself into a Job Shop (and) Flow shop marked distinctly.
- Considering that some of the resources (like DFPR, Solder bath etc.) will be shared between these 2 models, the Job shop will only have a limited Job shop layout – but any improvement is good enough to make some productivity gains.
- Going Forward All orders with order size < 6 will go through only Job Shop.
Given below is the product process matrix of the current as well as the recommended positions for the product.
Currently the company functions more as a job shop. This is amply evident due to three factors,
- Donner Company extensively supports midway design change from the customer, thus providing greater process flexibility to support the customer.
- It supports rush order which disrupt a more connected/continuous flow of work, thus reducing smooth flow of work as there would be in case of a more flow-oriented factory.
- The larger proportion of small orders (not in terms of no: of boards but orders) indicate that the company works on lower standardization and more of a job oriented operation.
The current recommendation for the company is to apportion its resources such that it works as a both a partial job shop and a partial process shop. This is indicated in the above product-process matrix.
-
For Orders > 8, which are deemed as “high risk” in nature owing to the nature of the technical complexity, layout design, innovation, customer’s profile etc. can be prototyped in the Job-shop before sent for mass production.
- The prototyping will be done in the Job shop and technical issues will be ironed out and then mass production will happen after customer’s acceptance in Flow Shop.
-
Based on the % distribution of orders, the approximate % of Labour Staff that needs to be assigned to the Job-Shop on a full-time basis is 17.25%. This includes about 2.3 heads on an average who are working on prototyping for the large orders.
- All Job Shop orders will be hand-held through by a single person right through the process (similar to Arthur Dief). This will lead to effectiveness and Quality in the Job shop.
- Because the order size < 6 is being done entirely in the Job shop, the Manual drilling machine can be moved there. It is not needed for the Flow Shop.
- Productivity Improvement
-
Follow the revised Organizational policies with respect to usage of CNC machines. This is in itself a big ticket item and will yield 30% productivity improvement – if the order profile of September was followed in the subsequent months.
-
The move of all order sizes < 8 to the Job shop will significantly reduce the Internal QC error rates in these orders and hence reduce the rework effort for the staff.
-
By following prototyping technique listed above, the quality improvement made will reduce the Field Error rate significantly. This will also save the rework effort.
- The job shop staff alone will need to spend time with the Customer Engineers to thrash out design issues and customer alignment issues. Hence the Flow-Shop staff can simply go-on and do a mass production of the orders based on the design. This will reduce the Average Inventory inside the Flow Shop and also reduce the time they waste on the telephone discussing Design and Delivery issues with the Customer Engineers.
The gains from this change are listed in Section-9 of this document.
- Quality Improvement
- The following changes discussed above will result in improvement in quality :-
- Small orders sent to Job-shop where single person will hand-hold the entire order through the entire standard production cycle. Although efficiency will be reduced in this model, the Quality will be very high.
- Prototyping will be done for all “high risk” orders and hence the rejects internally and externally will mainly be limited to low volumes. Hence when the flow-shop finally does the mass production of these capacities the reject rates are expected to be significantly lower.
- The Testing and Inspection activity can be strengthened by using improved tools like Probe to ensure the output is conformant to specification. They can also follow Six-Sigma, TQM and other Quality Systems in this activity to ensure they are able to catch more rejects internally, than from the field.
- Delivery Time Improvement
- As explained in Section-11 of this document, even with the increase sales target of 3M US$ annually, the delivery time of 15 days is alright and achievable because of the Productivity gains made.
- Additionally the changes being made to upstream processes (like procurement, sales, job scheduling etc.) should cut down on the cycle-time of order to delivery.
- The uneven workload in the 1st half of the month compared to 2nd half is a pure scheduling issue. Once the jobs are scheduled in an orderly fashion in Flow shop (with less interruption from small orders breaking the queue) the workload should get back to balance.
- Procurement
- The procurement team should not spend 2 days to research for suppliers and the lowest quote. They should proactively identify preferred suppliers and have an ongoing partnership contract for :-
- Delivery of Competitive Prices from time to time.
- Electronically Linked Supply Chain for Order Placement.
- Vendor Managed Inventory (VMI).
- Raw Materials (Epoxy sheets) to be delivered quickly to site with strict SLAs defined.
- The company should define an Inventory /Stock model for Epoxy sheets based on the contractual arrangement with the supplier. If it is not VMI model, then they should plan stocks for few days in advance.
- If it is possible instead of having strict turn-around SLAs with the supplier, they can offer the vendor to be part of Just-in-time (JIT) model.
- Job Scheduling
- The changes listed in previous sections should take care of uneven workload across the month, handling job delays etc.
- The Flow-shop will have staff mostly (not 100%) dedicated to the various departments or machines and will be able to do mass production with good quality.
- The revised Procurement Policy around Raw Material Inventory will enable the Production Manager to quickly commence production / schedule the order – much quicker than current timelines. This will reduce the over-all delivery time as well.
- Process Bottlenecks
- Donner has made a decision to live with Drilling as their bottle neck process. Because the order-size will now form the key of whether the order goes to Job shop or Process Shop, it mitigates the risk of the bottleneck. Also, the optimum usage of CNC drill and Manual drill should reduce the risk at this bottleneck.
- The Company has a limitation on the number of racks. It does not seem to be an item which needs to be a bottleneck. The company should stock sufficient volumes of this racks – so that it does not remain a bottleneck. As the company’s sales increases, the average Inventory is expected to release (refer Section-11) and hence there is a need to have sufficient number of racks.
- Sales Cycle & Future Readiness
- As explained in Section-11 of the document, the Flow-time based on the Order-size and the Cost per board are now available as a mathematical model. They should setup a process to review and revise these numbers from time to time. This will serve as an input to their Sales Process to be able to quote competitively and win new deals.
- As explained in Section-11 they are already ready to handle the load of 250K US$ sales per month. The productivity gains made in other areas have freed up available capacity to take up this additional work.
PGSEM2008 - Iimb.ernet.in Team – C1 Page