Control Phase

The project has continued and is beating our expectations. The team contributed very well and we believe that the project has far exceeded our expectations. The Finance department have also come back with a batch improvement valuation. The savings per batch is now calculated at $2056 per batch. There is a cost included here for testing which they were not including at the start of the project but when it was reviewed showed that it was a realistic and large saving to the laboratory department.

Since the 2 failures we have not proceeded with the timing process unless the particle size analysis after application has been complete. The particle size analysis has so far indicated that our D90 should be below 1450 for the processing at 115 TC. As can be seen from the run chart below the results have been in a tight range.

Run chart

We have manufactured 42 batches since the failure and we have noticed considerable improvement on the CP/CPK. Before the project started the CPK was a negative number and now it is positive (CPK 1.61). This can be seen from the chart below.

Process capability chart

If we include the 2 failed batches the CP/CPK is as follows 1.18/1.17 and while both are positive values they are both below the 1.33 minimum expectations.

Process capability chart 2

Our target was to make reductions to the lead time, number of days timing and the cost per timing at the timing stage. The table below shows the current state. Each metric has exceeded our expectations.

Lead time: This has now been adjusted on SAP to reflect 30 days. However with machine and personnel availability the process can be complete in 8 days

Number of days timing: None of the last 42 batches manufactured required day 2 timing so this is exceeding expectations as we had hoped to be 1.5 days timing or less. On SAP we have removed all phases after day 2 timing and we will monitor and review this in 6 months’ time with a view to only leaving 1 timing phase for day 1 only.

Cost per batch at the timing stage: This cost has reduced from a massive $2893 to $847. We had been expecting a smaller figure as we were not getting credit for a considerable reduction in testing which can and actually does add a large cost to each day of timing. Our Finance department reviewed the potential savings and have agreed that there was a large work reduction for the laboratory and have now removed that from the cost of manufacture of each batch. The order for the full year 2013 was expected to be 100 batches but since the start of the project the order has increased to just over 200 batches.

Table with previous and current state

Metric	Previous state	Current State
Lead time Number of days timing Cost per batch at the timing stage	42 days 2.2 days $2893 per batch	30 days 1 day (last 42 batches) $837

We have made a huge improvement in the stages of processing of this product. We have not changed the formulation of the product but with the help of the DMAIC process we were able to show that the product can be manufactured more efficiently and have a better use of equipment and personnel resources. We will now proceed with an OEE on the equipment as we have shown that the product can be manufactured in 8 days but we will need to show that the equipment effectiveness is in place before we reduce the lead time further. A lead time of 8 days on this product would mean that a customer could have their product 2 months after they have ordered it.

We will present this project to the site leadership team and already I have been asked to apply these learning’s to other products.

Tasks completed

·        Changed from 75 timing coats on day 1 to 115 timing coats

·        Updated MBR with instruction to remove and test a 10 gram sample for particle size

·        Test the sample using the EYECON which is now qualified

·        SOP implemented to cover use and cleaning of the EYECON

·        Updated the SOP for the process guidelines of the product

·        Run chart for the product and the bead blends has been started and will be present as a KPI at the monthly communication meetings for a period of 1 year
Below is the equipment used in the processing at the timing stage.
The CF Granulator

 

Improve phase

The team have been working well and we are on schedule with our project. We agreed on 115 timing coats per batch on day 1 based on the review of the 4 batches done under VCA.

Our validation department have requested that we track the next 8 batches under VCA and that we submit a manufacturing engineering report when it is completed. We have also to manufacture these batches under process deviation as we are making changes to the batch record. The changes are not implemented yet.

We successfully processed the 8 under VCA. We then proceeded with another 20 batches and hit a Road block. We had 2 out of spec batches as can be seen from the graph below.

6^th hour dissolution Run chart

We called an emergency meeting to root cause analyse and we performed a fish bone. All the manufacturing associates that were involved in processing the batches were interviewed. The laboratory personnel that tested the batch were also interviewed.

Fish bone

Both the manufacturing associate and the laboratory person that tested the batch had observed that the bead was bigger than normal. It was now time to check our particle size after the applied stage as we had been removing a 10 gram sample for analysis with the EYECON particle size analyser The samples were not analysed yet so we proceeded to perform the analysis on the batches. The table below shows the results of the 28 batches analysed and to our surprise there were 2 batches with larger bead than the other 26 batches and both were outside specification on dissolution.

Particle size analysis

What caused the larger bead?  It was time to go back to Gemba and check the CF room and Compu 4 system. We reviewed the data on the batch report and found that a critical parameter (the rotor speed had not been adjusted when it should have been. The associate that had manufactured the batch had not increased the rotor speed at the required time and this caused the bead to get larger early in the batch which had the affect on the sixe of the bead at the final stage of processing.

I had to prepare slides for our Product review board and state the case for not stopping the project and reverting back to 75 timing coats at day 1. Before the presentation there was a lot of doubt but at the end of the presentation it was agreed to go ahead.

The project has still remained on track despite delays in implementing changes on our batch record and SOP’s which are currently in change control. We are now keeping a closer watch on the particle size until the EYECON has been qualified which is currently under way after a recommendation from the Product Review Board. The Product review board also recommended that we immediately but the capsule batches using these components on for stability purposes to ensure that the stability of the batches are not affected.

Gant Chart

With the exception of the 2 batches that are outside specification the other 26 batches processed at 115 timing coats have met the acceptance criteria. They will not require further timing as the profiles are suitable to use in the bead blends with our current safety stock of components. We will now continue to monitor the batches and evaluate the savings being made for the product.

Regression analysis was not applicable as there was no change in the processing parameters. We did do a Regression analysis just to see what appeared but there was nothing of benefit.

 

Analyse Phase

E-log book entry for Analyse stage ( Q24h process improvement)

The project is progressing well and we have been meeting twice a week with good interaction between the team and good participation.

We performed a risk assessment that looked at changing the number of timing coats on day 1 from 75 timing coats to a higher number. The risk assessment looked at the absence of I days drying and what impact this could have on the assay and dissolution profiles.

We had a project that had been performed on the drying of batches in ovens previously called the Emerald project and it showed that even though the batches were dried for 20 hours, it had little impact after 4 hours as the purpose of the drying was to take off excess solvent and moisture. The data analysis here showed that there was no risk present.

We also assessed the chances of overtiming the batches. Again the analysis of the data showed that 1 batch of the previous 100 batches was in spec at 95 timing coats but if we gave it another 20 timing coats the batch would still be above the lower spec of 20.

Based on the risk assessment we decided to go with 100 Timing coats for 2 batches and 110 timing coats for 2 batches.

Our manufacturing associate processed the 4 batches at the batches were sampled and sent to the labs for testing. The results were as follows with 100 TC on batches at the upper limit at 59 and 62 and the 2 batches at 110 TC at 53 and 45.

When we look at where the batches have shifted to, we were very happy with our progress. While some of the team were disappointed that the batches could not be used in a bead blend after day 1 timing I was relieved that the results were not much lower as our previous review was indicating 115 TC as the target value.

On further review of our components in stock we decided to call the batch at 45 in spec. the team were now satisfied, as we were starting to show a financial benefit to our project.

The table below shows the run chart including the increased number of timing coats. After processing the 4 batches we reverted to 75 TC while awaiting the results and those batches are also shown on the run chart.

We decided to give the 3 of the 4 batches further timing up to 120 TC as this would give us confidence in the increased number of timing coats.

The 4 batches in the chart below can be seen in a tight range in the middle of the limits after day 2 timing. During our analysis of the data we could calculate that each timing coat would move the dissolution by 0.8 and therefore indicated that we could have given the batches 140 TC and still not have an out of spec batch. The lowest projected at 140 TC would be 21.

Our confidence is growing and the expectation is high. Early indications are that with a higher number than 100 TC would have all batches in spec although they may not be usable in bead blends without further timing for some.

Although these 4 batches were not processed as a design of experiments (DOE) we gained knowledge in this regard. We have performed DOE on our products in the past but after consultation with our Quality and Validation departments and the risk assessment it was agreed that a DOE would not give us any greater information on the number of timing coats as we were not changing any of the processing parameters.

Our processing parameters included

Slit air voume, slit air temperature, Product temperature, rotor speed, Talc flow rates, Solution flow rates, gun solution pressure and dewpoint.

We were happy with the target values of the parameters in the batch record. The CF is run by a COMPU 4 computer system that is recording a large number of variables that are not recorded on the batch record for this process but are available for reference if needed.

Measure Phase

E-log book entry for Measure stage
 We have been meeting regularly and the team is functioning well. The manufacturing associates have been gathering data by reviewing over 100 batch records.
They gathered information like bin loss, hopper loss and application yields.

We also gathered information about particle size from the batches being processed at the moment. We have a new piece of equipment on site for analysing particle size but it is not qualified yet.

  To ensure that the equipment is reliable we have taken samples and analysed the samples with the new equipment and the sieve shaker used in the laboratory. The results were surprisingly very close to each other.
We continue to use the equipment but only for information purposes. The team don’t think particle size is important but I feel that it could be valuable information.
We also did a process capability on the batches and a run chart to show our progress.

The 2 charts show that the results are in a tight range but above the upper spec limit. A shift of these results downwards could have the results in the centre of the spec limits. We have reviewed the typical number of timing coats required to get the desired result.

With the help of our statistician, we did a gauge R&R study  
The table indicates that there is a statistically significant difference between the operators but no significant difference between parts (individual results) or operators by parts (the expected variances and standard deviations are almost zero).


The indications from our measure phase are that 115 TC should be the number of timing coats to use to get the desired profile.

 However we will perform a risk assessment and after that we will determine the route to go. We have spoken to our Quality and Validation departments and they have confirmed that any changes made will need to be monitored under a VCA (validation corrective action).

The member of our team that deals with compliance has reviewed the current regulatory filings and we do have scope for introducing process improvements.

Define

I am a Sligo IT student on the second year of a BSc in manufacturing management. As part of my six sigma DMAIC project I have chosen the Q24h process improvement in the organisation that I work. The product is being manufactured in the company for over 20 years and is  labour intensive and has a large lead time. I put forward a proposal to reduce the number of days timing and also the lead time.

The project proposal was accepted by our operational excellence black belt review team and the Finance department. The Finance department evaluate the proposal to ensure it meets the criteria for the particular project.

My project will go through all the phases of DefineMeasureAnalyseImproveControl
Define
We have started the define phase by gathering together a diverse project team which consists of a person from Quality control, Manufacturing processing, Manufacturing compliance, SAP specialist, paperwork coordinator, Process engineer(statistician)   and Validation department.
The team

Process map
In our first meeting we decided that it was important to do a map of the Q24h process. The process is an intricate process and it would be helpful in achieving our goal.

Brain storming
The team also felt it would be an important exercise to brain storm. I was very close to the process and this may affect our judgements. During the brainstorming every ones opinion counts and put forward as many ideas as possible. The brainstorming activity suggested that we should increase the number of timing coats applied on day 1 and this is what our project will focus on.

SIPOC
This was another tool we used in the define phase and it was important to have a look at the Supplier Input Process Output Customer