Recently, some of you may have seen tweets or updates about the Littlefield sim. The Littlefield sim is a simulated factory which runs in accelerated real time. Each sim-day lasts one hour of real time. MBA students at various schools across the country often compete within their class groups or school groups to run the factory in the most efficient and profitable manner.
There are many ways that Littlefield can be configured. For my LFGSM Operations Management class, we played the sim twice. The first instance was a “light” version that focused specifically on planning capacity. We were given 50 days of data to analyse prior to the sim start. The sim would then run for 7 real life days. At the end of simulated day 218, the computer would take over and run for another 50 days without our interaction. There were several intricacies to figuring out a strategy. My team agreed early that, for this light version, we wanted to do very little day-to-day monitoring. As we started with $2M in cash, with machines costing between $80K and $100K, we decided to buy all the machines we would need on the first day. To do that, we had to estimate demand from the first 50 days. The sim instructions also advised that the simulated demand would roughly follow the standard product life cycle (linear increase, then flatline, then linear decrease). Using the first 50 days and a linear (least squares) regression, DH and I estimated the eventual peak average demand and planned machines accordingly. (DH acted as a data analysis consultant for me, and he did a great job. 🙂 ) I provided this input back to my team, and we bought machines on the first day. After that, we basically had nothing to do. Late in the game, a spike in demand occurred (in line with our forecasted averages but still much higher than the standard deviation) which overloaded our capacity. Several teams erred by buying a machine at the late stage. I quickly devised a cost-benefit analysis to determine if buying a machine made sense, and it did not. My team trusted my analysis. We ended up winning the first sim thanks largely to the forecast accuracy and the cost-benefit analysis.
Because of my success in the first one, my team put me largely in charge for the second instance of Littlefield. For the second instance, there were many more levers to pull. We could choose any of three pricing contracts to pursue. We had to manage inventory purchases and supply, which was automatically managed by the sim in the light version. We needed to purchase machines but started with only $70K, which was barely enough for an initial inventory order, much less machine purchases. Also, instead of only 50 days of automated play at the end, this version had 100 days, and leftover inventory and machines on the last day would not be sold back. Demand would follow an average rather than the product life cycle, making forecasting mostly irrelevant. Fortunately, superior data analysis paved the way again, combined with diligent attention to the sim. For this variant, our team had agreed on 24-7 monitoring. We each took a 24 hour shift, with a split of 12-hour shifts at the point that we’d have to start repeating. I was the only person who did a single 24-hour shift and no 12-hour shift, but I also did the most pre-sim prep work on data analysis and strategy. We started with a very risk averse plan, wherein we required a good bit of collaboration and verification before shifting to the highest price contract. For machines, I’d built a cost-benefit calculator based on the analysis I’d done from the previous sim. For re-order point (ROP) and order quantity (ROQ), another member of our team had pulled together a calculator based on the formulas in our text. We also had quite a few areas to monitor with things to tweak for certain contingencies.
About halfway through the sim, we’d bought all the machines we needed, and I was puzzling over why the ROQ was so low. It seemed to me that, given the fixed cost of ordering inventory ($1K per order) and given that we had very little to do with the cash we had on hand, we should just spend all the cash we had on inventory, so long as the inventory total was less than the lost interest we’d earn on the cash. Yet, the ROQ formula didn’t come out to that value. DH was nearby, so I asked him to help me take a look. He had me explain the formula to him, and in trying to explain it to him, we stumbled on the problem. The denominator had been set to the cost of each unit rather than the holding cost of each unit. With that adjustment, the ROQ increased tenfold to line up with my own scenarios for break-even on interest vs. inventory. I dashed off an e-mail to the rest of the team explaining and prepared to order inventory to use up all available cash. After that, it just became a matter of closely watching for when we neared a ROP to set ROQ appropriately.
Another piece was the price contract. Though we started conservatively, we started bumping up to $1250 contracts periodically once we were at full capacity. However, we kept switching back to $1000 contracts when lead times rose due to higher-than-average demand. For the final settings, I ended up doing an analysis of the past 95 sim-days and discovered that if we’d simply left it set to $1250, we’d have been $20K richer. Basically, so long as you’re hitting the appropriate lead times at least 80% of the time, you’re better off leaving it set to $1250 rather than switching back and forth and thus potentially losing some $1250’s to $1000’s. So, that was one mistake we made.
I ended up taking over the sim for the last sim-day before the computer automation takeover. Since I knew the formulas the best, the team asked me to figure out that last day’s orders. I ended up forcing a big material order on the last day and then setting inventory to basically be just-in-time (JIT) ordering, with ROQ equal to ROP. That basically minimized the amount of inventory we’d have on hand for the final day.
I won’t know until the final class of the term on Thursday evening, but I think that big last order is what pushed us into the lead versus the other teams. We won again (we’re told that is rare for this sim, for the same team to win on both, though I don’t know why), which is awesome. I was in nervous knots waiting on the final result on Friday evening. I get that way for big milestones at work, but not to the same degree that I was wrapped up in it for this project. I cannot imagine what it will be like when/if I’m actually running a company! I hope that, by then, they’ll have invented better acid reduction drugs. 🙂
Note: If you’re a student doing Littlefield who has come across this information via Google, good luck with your simulation. I will not answer questions about strategy nor will I provide our notes, spreadsheet, or other data. What’s in this post is all you get… 🙂